• Re: G.722

    From Marco Moock@mm@dorfdsl.de to uk.telecom on Fri Mar 20 17:12:25 2026
    From Newsgroup: uk.telecom

    On 20.03.2026 15:22 Uhr Richmond wrote:

    Marco Moock <mm@dorfdsl.de> writes:

    On 20.03.2026 10:48 Uhr Richmond wrote:

    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it
    considers quality and latency. Progress seems slow. Signal has
    a better frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't
    have any advantage.

    Why does the ISP need to do processing?

    If the 2 subscribers use different codes, the signal needs to be
    converted.

    But they can negotiate the same codes.

    Only if the end devices negotiate, but in most cases they negotiate to
    the SIP provider and that relays the call to the destination or transit providers.
    --
    kind regards
    Marco

    Send spam to 1774016531muell@stinkedores.dorfdsl.de

    --- Synchronet 3.21e-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Mar 20 17:15:48 2026
    From Newsgroup: uk.telecom

    On 20/03/2026 10:48, Richmond wrote:
    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it considers
    quality and latency. Progress seems slow. Signal has a better
    frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't have
    any advantage.

    Why does the ISP need to do processing?

    Well no more than any other packets being routed, and bandwidth is low...

    AIUI the SIP server you are (permanently) connected to takes incoming
    packets and routes them to you. The permanent connection thus bypassing
    any NAT you have in place

    It 'knows' about other SIP servers and routres accordingly. I dunno how.

    This sort of thing seems insignificant to me, compared to streaming
    audio and video.

    Of course it is.

    The problems I think are NAT, Firewalls, and publicity. If you know
    someone's SIP address you ought to be able to make a direct call over
    the internet in the same way as Signal does. It would be free of any per minute cost. The compression would be done with equipment at either end,
    not by the ISP. That's how it looks to me anyway.

    You need the server to act as a proxy to circumvent the NAT.
    --
    Future generations will wonder in bemused amazement that the early twenty-first centuryrCOs developed world went into hysterical panic over a globally average temperature increase of a few tenths of a degree, and,
    on the basis of gross exaggerations of highly uncertain computer
    projections combined into implausible chains of inference, proceeded to contemplate a rollback of the industrial age.

    Richard Lindzen

    --- Synchronet 3.21e-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Mar 20 17:16:50 2026
    From Newsgroup: uk.telecom

    On 20/03/2026 11:58, Andy Burns wrote:
    Richmond wrote:

    If you know someone's SIP address you ought to be able to make a
    direct call over the internet in the same way as Signal does. It
    would be free of any per minute cost. The compression would be done
    with equipment at either end, not by the ISP.

    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather than
    via the provider.

    Are you sure? I dont think that would bypass NAT...

    But I don't think you'd want a tidal wave of AI vishing calls would you?
    --
    There is something fascinating about science. One gets such wholesale
    returns of conjecture out of such a trifling investment of fact.

    Mark Twain

    --- Synchronet 3.21e-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Mar 20 17:17:48 2026
    From Newsgroup: uk.telecom

    On 20/03/2026 15:03, Marco Moock wrote:
    On 20.03.2026 11:58 Uhr Andy Burns wrote:

    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather
    than via the provider.

    I haven't seen that. Are there SIP setups that handle it this way?

    I have set up my SIP service using IPv6 and have not opened ports on
    IPv4, but I can do phone calls to people with only IPv4, so there
    cannot be a direct connection in that case.

    +1. there is not. The SIP server is a proxy
    --
    There is something fascinating about science. One gets such wholesale
    returns of conjecture out of such a trifling investment of fact.

    Mark Twain

    --- Synchronet 3.21e-Linux NewsLink 1.2
  • From Andy Burns@usenet@andyburns.uk to uk.telecom on Fri Mar 20 18:29:21 2026
    From Newsgroup: uk.telecom

    The Natural Philosopher wrote:

    Are you sure? I dont think that would bypass NAT...

    The system I was tracing has a session border controller, with public
    IPs non-NATed,
    --- Synchronet 3.21e-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Sat Mar 21 14:00:23 2026
    From Newsgroup: uk.telecom

    On 2026/3/20 11:58:39, Andy Burns wrote:
    []
    But I don't think you'd want a tidal wave of AI vishing calls would you?

    Fishing, I used to know. Phishing, I sort of know these days. but what's vishing - video phishing? Something else?
    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    Odds are, the phrase "It's none of my business" will be followed by "but".
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Sat Mar 21 14:12:51 2026
    From Newsgroup: uk.telecom

    "J. P. Gilliver" <G6JPG@255soft.uk> writes:

    On 2026/3/20 11:58:39, Andy Burns wrote:
    []
    But I don't think you'd want a tidal wave of AI vishing calls would you?

    Fishing, I used to know. Phishing, I sort of know these days. but what's vishing - video phishing? Something else?

    Assuming it is Voice Phishing, I think the solution is the same as for
    SPAM and Cold calling, i.e. a whitelist of some sort. It's not a great
    solution but it will do while I wait for something else. I have to
    switch it off if I am expecting a call from an unknown number but I can
    switch it on again afterwards.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Andy Burns@usenet@andyburns.uk to uk.telecom on Sat Mar 21 15:23:21 2026
    From Newsgroup: uk.telecom

    Richmond wrote:

    J. P. Gilliver writes:

    Andy Burns wrote:
    I don't think you'd want a tidal wave of AI vishing calls would you?

    Fishing, I used to know. Phishing, I sort of know these days. but what's
    vishing - video phishing? Something else?

    Assuming it is Voice Phishing
    Correct; another term for it would be SPIT (spam over internet telephony).

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Wed Apr 1 16:24:35 2026
    From Newsgroup: uk.telecom

    On 17/03/2026 22:45, Richmond wrote:
    I was investigating whether using a proper SIP phone, as opposed to a
    DECT phone plugged into an ATA, would provide a better sound quality, including the lower pitch sounds.

    G.722 has the same coverage of low pitched sounds as G.711. It's
    advantage is in handling higher pitched sounds, at the cost of more quantisation noise.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Wed Apr 1 16:31:46 2026
    From Newsgroup: uk.telecom

    On 18/03/2026 21:32, Richmond wrote:
    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol) doesn't necessarily choose the highest level, because it considers quality and latency.

    SDP doesn't negotiate anything. It simply lets each side say what it is prepared to receive. The side making the second offer is encouraged to
    only offer codecs offered in the first offer, and even to prefer the
    first acceptable offer, but that is ultimately a decision for the user
    agents. The user agents are also encouraged to actually use the same
    codec each way, but that isn't something mandated by SDP. It is very
    common for two codecs to be used, one for normal speech and one for
    DTMF, in the same call.

    The only way the side making the first offer can reject the second offer
    is by aborting the call (or in the case of late offer, by immediately
    ending the call.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Wed Apr 1 16:43:01 2026
    From Newsgroup: uk.telecom

    On 20/03/2026 11:58, Andy Burns wrote:

    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather than
    via the provider.

    Whilst that is possible, I don't think it is common, and, as others have implied, it would be a nightmare with NAT, let alone the CGNAT that a
    consume, and, in particular, mobile phone user, is likely to encounter. Historically, one side would have been ISDN (circuit switched), so it wouldn't have been possible.

    At least in the USA, I think you can get a halfway house, where a retail
    ITSP offers the media address of its wholesale ITSP.

    (Asterisk, an open source SIP PABX (more correctly toolkit) only
    officially supports setting up media on a one hop basis, and then reconfiguring the media path to cut out intermediates.)

    Richmond wrote:
    The compression would be done
    with equipment at either end, not by the ISP.

    Codecs are converting between analogue and digital domains, not doing
    simple compression. A 16 bit, 44.1kHz codec would still be a codec,
    even though it is CD quality.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Wed Apr 1 16:56:30 2026
    From Newsgroup: uk.telecom

    David Woolley <david@ex.djwhome.demon.invalid> writes:

    On 20/03/2026 11:58, Andy Burns wrote:
    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather
    than via the provider.

    Whilst that is possible, I don't think it is common, and, as others
    have implied, it would be a nightmare with NAT, let alone the CGNAT
    that a consume, and, in particular, mobile phone user, is likely to encounter. Historically, one side would have been ISDN (circuit
    switched), so it wouldn't have been possible.

    At least in the USA, I think you can get a halfway house, where a
    retail ITSP offers the media address of its wholesale ITSP.

    (Asterisk, an open source SIP PABX (more correctly toolkit) only
    officially supports setting up media on a one hop basis, and then reconfiguring the media path to cut out intermediates.)

    Richmond wrote:
    The compression would be done
    with equipment at either end, not by the ISP.

    Codecs are converting between analogue and digital domains, not doing
    simple compression. A 16 bit, 44.1kHz codec would still be a codec,
    even though it is CD quality.

    I don't know the context of that quotation of mine above, but, codecs do compression right? mp3 does compression, mp3 is a codec? They don't
    *have* to do compression, but if you want to get more data over an old
    copper phone line you would do compression.

    If you have two VOIP customers speaking directly to each other, can you
    not get better sound quality than analog? This must be possible I think,
    as you can stream movies over a copper wire, so improving the sound
    quality of voice ought to be a walk in the park.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Wed Apr 1 20:43:38 2026
    From Newsgroup: uk.telecom

    On 01/04/2026 16:56, Richmond wrote:
    I don't know the context of that quotation of mine above, but, codecs do compression right? mp3 does compression, mp3 is a codec? They don't
    *have* to do compression, but if you want to get more data over an old copper phone line you would do compression.

    It is conventional to use 'codec' more broadly to encompass compressed formats.

    If you have two VOIP customers speaking directly to each other, can you
    not get better sound quality than analog? This must be possible I think,
    as you can stream movies over a copper wire, so improving the sound
    quality of voice ought to be a walk in the park.

    When I was using a VOIP linme and an analogue line into te same PABX,
    the VOIP qualiy was superior.

    Remember uncompressed analogue phone data is 64kbit. Its possible to
    sample much higher than that and use compression to get bit rates down,
    Speech in particular is rather lacking in the sort of information a full orchestra has, so compresses very well.
    --
    "It was a lot more fun being 20 in the 70's that it is being 70 in the 20's" Joew Walsh

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Wed Apr 1 23:54:29 2026
    From Newsgroup: uk.telecom

    On 01/04/2026 16:56, Richmond wrote:
    I don't know the context of that quotation of mine above, but, codecs do compression right? mp3 does compression, mp3 is a codec? They don't
    *have* to do compression, but if you want to get more data over an old

    That's an implementation detail, and assumes that the initial A/D to conversion produces a large number of bits, which is probably true.

    copper phone line you would do compression.

    That a different topic. In the days of voiceband modems, compression
    was used, towards the end of their reign, but most high volume data is
    is already in a format that won't compress further, so I think, since broadband, it has not been done routinely.

    They way they get lots of data down a twisted pair transmission line (aluminium would be similar to copper, and silver might be better) is
    actually to send thousands of low speed data streams (4,000+ or 8,000+
    for VDSL variants). The idea of using lots of sub-carrier frequencies is something of a analogue concept, although it will actually be
    implemented digitally, these days. Different sub-carriers will be
    impacted differently by the imperfections in the transmission line, but
    there won't be enough variation within a sub-carrier to invalidate it completely.

    They also actually expand the data, because they use error correcting
    codes, which can, for example fill in for a sub-carrier that fails to
    work. For VDSL, the expansion is about 8%.

    Regarding another reply, "codec" is often used to mean the output format produced by a codec, rather than the the actual process that does the
    coding and decoding, but the word "codec" is a combination of code and
    decode, so refers to the process.


    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Thu Apr 2 00:31:57 2026
    From Newsgroup: uk.telecom

    On 01/04/2026 16:56, Richmond wrote:
    If you have two VOIP customers speaking directly to each other, can you
    not get better sound quality than analog?

    In a literal sense, no, because human voice generation and hearing have
    major analogue components, and the transducers used (microphones and
    speakers) work with analogue signals. However you really meant analogue
    to the premises phone lines, which means 3.1kHz audio service (3.1kHz bandwidth from 300Hz to 3.4kHz). If you have unlimited data rates, you
    can produce arbitrarily high quality sound with digital systems, but the analogue phone constraint you are applying fixes audio bandwidth.

    Analogue circuits aren't limited to 3.1kHz audio. Before the world went digital, for example, those used to distribute FM broadcast material to transmitters had a lot more audio bandwidth than phones, and even those
    used for AM radio distribution would have needed more bandwidth than
    3.1kHz phone connections.

    It should also be pointed out that nearly all analogue line phone calls
    are converted into digital, using a G.711 codec, so you don't, in
    practice, get analogue end to end calls, except with toy telephones. I
    think I've heard that G.722 is used for some circuits, in some
    countries. (Note that G.711 is often referred to by corrupted versions
    of the names of two common variants, A-law (typically Europe) and -|-Law (America and Japan). You will often see -|-Law referred to as ulaw,
    because people confuse -| with u, when it is actually pronounced like m.
    A-law is less corrupted, but often appears as alaw.)

    End to end analogue signals degrade as they travel. Digital signals
    also degrade in the analogue forms in which they are modulated, but can
    nearly always be regenerated to their original form, although, if that
    fails, the result can much severe than the degradation of an analogue
    signal.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Thu Apr 2 01:08:06 2026
    From Newsgroup: uk.telecom

    On 2026/4/1 23:54:29, David Woolley wrote:
    []
    Regarding another reply, "codec" is often used to mean the output format produced by a codec, rather than the the actual process that does the
    coding and decoding, but the word "codec" is a combination of code and decode, so refers to the process.

    I thought it was a combination of coder and decoder, i. e. it's the
    actual hardware (or software, or firmware).

    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Thu Apr 2 01:14:06 2026
    From Newsgroup: uk.telecom

    On 2026/4/2 0:31:57, David Woolley wrote:
    []
    End to end analogue signals degrade as they travel. Digital signals
    also degrade in the analogue forms in which they are modulated, but can nearly always be regenerated to their original form, although, if that

    In that "digital signals" always, for most of about the last
    fortysomething years, means "digital signals with error correction";
    this is so often assumed as to be understood, but the omission of the
    fact confuses newcomers to the subject (who are left wondering _why_ are digital signals better - and/or more reliable - until this point is
    explained. Which it often isn't).

    fails, the result can much severe than the degradation of an analogue signal.

    Oh yes, the "digital cliff", where the error correction fails. Although
    the improvement in going to (corrected!) digital is vast (and gives me
    many more channels), I find it _much_ more irritating when it _does_
    fail - for video - than a noisy analogue channel. I think it's because
    there is so much less warning, so it's so sudden.
    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Thu Apr 2 10:13:45 2026
    From Newsgroup: uk.telecom

    David Woolley <david@ex.djwhome.demon.invalid> writes:

    On 01/04/2026 16:56, Richmond wrote:
    I don't know the context of that quotation of mine above, but, codecs do
    compression right? mp3 does compression, mp3 is a codec? They don't
    *have* to do compression, but if you want to get more data over an old

    That's an implementation detail, and assumes that the initial A/D to conversion produces a large number of bits, which is probably true.


    As I understand it G.722 does do some compression to allow it to have a frequency range going down to 50Hz, whereas the analog phone only goes
    down to 300Hz. This latter difference makes the voice sound thin and
    distant.

    Google Meet on the other hand has a much fuller sound with those low frequencies making it sound more realistic.

    The compression G.722 uses is ADPCM (Adaptive Differential Pulse Code Modulation).

    https://en.wikipedia.org/wiki/Adaptive_differential_pulse-code_modulation

    So really the question is, with the right SIP/VOIP telephone equipment,
    and a VOIP to VOIP connection, can I achieve this full frequency? I do
    have FTTP. It seems the things standing in the way are legacy, NAT, and firewalls.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Thu Apr 2 10:43:55 2026
    From Newsgroup: uk.telecom

    I found this:

    https://www.vodafone.co.uk/newscentre/press-release/vodafone-brings-hd-voice-call-quality-to-customers/

    "Vodafone UK has rolled out High Definition Voice technology across the UK Customers making Vodafone-to-Vodafone calls on compatible phones will
    enjoy exceptionally clear conversations with family, friends and work colleagues"

    Unfortunately their customer service is intolerable to me.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 13:19:03 2026
    From Newsgroup: uk.telecom

    On 01/04/2026 23:54, David Woolley wrote:
    They way they get lots of data down a twisted pair transmission line (aluminium would be similar to copper, and silver might be better)

    When I put the internet into the channel islands, Guernsey Telecom
    laughed at Jersey Telecom who had used Aluminium.

    While Guernsey could get decent ADSL to all their customers from the
    single exchange, attenuation was too high for Jersey's aluminium
    connected customers...
    --
    No Apple devices were knowingly used in the preparation of this post.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 13:53:48 2026
    From Newsgroup: uk.telecom

    On 01/04/2026 23:54, David Woolley wrote:
    The idea of using lots of sub-carrier frequencies is something of a
    analogue concept, although it will actually be implemented digitally,
    these days.-a Different sub-carriers will be impacted differently by the imperfections in the transmission line, but there won't be enough
    variation within a sub-carrier to invalidate it completely.

    In fact they use what are referred to as 'bins'. Each bin represents a
    single carrier frequency in the range IIRC of 50Khz to around 2,5Mhz -
    but don't quote me on that,
    and as part of the probing the quality of each bin, and hence the number
    of bits it can accommodate, is determined.

    All comms is a (digital) signal imposed on an analogue carrier. Unless
    you are counting atoms and electrons individually *everything* is
    analogue at the bottom.

    I cannot recall the exact modulation schemata used on ADSL but its
    carefully designed by engineers with a more mathematical bent than I
    ever had, to do the best possible under the constraints of the
    transmission media they have to use.

    DSL over copper pair is subject to massive interference from long medium
    and short wave radio and transmits in that band itself, and cross talk
    between pairs also limits performance, As do intermodulation products
    created by partially oxidised contacts forming crude 'metal oxide
    rectifier' type diodes.

    Running VOIP over the digital signals carried by DSL doesn't really have
    any quality issues related to that layer unless packet loss becomes too
    high or bandwidth becomes too low. Mobile phone audio degradation is a function of available bandwidth being lowered to astonishing levels of crappiness, it being judged that a degraded signal is better than no
    signal at all. None of that should apply to VOIP over 'broadband'.

    In general the old baseband phones were converted to 64kbps synchronous
    TDM with iirc an 8 bit 8kHz sample rate giving an effective upper
    frequency limit of about 3Khz and S/N ratio of about 45dB. I*n those
    days (1960s?)the computer power to do compression didn't exist and nor
    did the algorithms

    It's very easy to do a lot better than that with a compressed VOIP
    signal, and in today's broadband world you can go a LOT faster than
    64kbps if you want. And voice, with distinct lack of much at the top
    end of the frequency band and plenty of pauses, and only a single
    frequency set, is absolutely ideal for compression.

    I heard once (apocryphal) that intelligible speech could be transmitted
    at 50 bytes per second.

    G722 is actually slightly above the old 8kHz sample and the bandwidth
    extends to 7kHz. And does use compression because the final bitrate is variable at 16, 24 and 32 kbps

    I have no idea if it is the 'general standard' but one things is clear,
    that excluding *mobile* (RF) phone technology, VOIP over broadband is
    superior to a baseband phone in every respect.
    --
    Any fool can believe in principles - and most of them do!



    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 14:00:24 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 01:08, J. P. Gilliver wrote:
    On 2026/4/1 23:54:29, David Woolley wrote:
    []
    Regarding another reply, "codec" is often used to mean the output format
    produced by a codec, rather than the the actual process that does the
    coding and decoding, but the word "codec" is a combination of code and
    decode, so refers to the process.

    I thought it was a combination of coder and decoder, i. e. it's the
    actual hardware (or software, or firmware).

    Of course it is. Certain people don't know as much as they want you to
    believe they do,

    Codec is a device that interfaces between generally analogue signals of
    some sort at one end and a digital data stream at the other. How that
    digital data stream is transferred is implementation specific. As is
    any signal processing, digital or analogue, that is carried out by the aforesaid codec.

    It's mostly applied to software layers these days with the actual
    digital to analogue (and vice versa) being performed by devices with
    other names, like 'modem' or 'DAC'
    --
    ThererCOs a mighty big difference between good, sound reasons and reasons
    that sound good.

    Burton Hillis (William Vaughn, American columnist)

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 14:11:56 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 10:13, Richmond wrote:
    David Woolley <david@ex.djwhome.demon.invalid> writes:

    On 01/04/2026 16:56, Richmond wrote:
    I don't know the context of that quotation of mine above, but,
    codecs do compression right? mp3 does compression, mp3 is a
    codec? They don't *have* to do compression, but if you want to
    get more data over an old

    That's an implementation detail, and assumes that the initial A/D
    to conversion produces a large number of bits, which is probably
    true.


    As I understand it G.722 does do some compression to allow it to
    have a frequency range going down to 50Hz, whereas the analog phone
    only goes down to 300Hz. This latter difference makes the voice sound
    thin and distant.

    The bottom end frequency range is not a function of compression

    Google Meet on the other hand has a much fuller sound with those low
    frequencies making it sound more realistic.

    The compression G.722 uses is ADPCM (Adaptive Differential Pulse Code Modulation).

    Ah. Didn't know that. simple and very effective for lower frequency stuff.

    https://en.wikipedia.org/wiki/Adaptive_differential_pulse-code_modulation


    So really the question is, with the right SIP/VOIP telephone
    equipment, and a VOIP to VOIP connection, can I achieve this full
    frequency?

    Yes.

    I do have FTTP. It seems the things standing in the way are legacy,
    NAT, and firewalls.

    None of those are significant if you are using a proxy server which is
    how all systems must work, precisely because of NAT and firewalls.
    NAT allows *you* to open and maintain a connection but effectively
    denies incoming attempts to connect.

    By maintaining a *permanent* connection to a SIP gateway, that
    restriction is overcome. At any time an incoming call will have a NAT
    port and IP address to connect to, provided the data is coming from the
    SIP gateway.

    So you cannot in legacy IPV4/NAT land, call another internet phone
    directly.,
    Hence the need for a SIP server.
    (Or a Whatsapp, Skype etc server).
    --
    I would rather have questions that cannot be answered...
    ...than to have answers that cannot be questioned

    Richard Feynman



    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 14:14:22 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 01:14, J. P. Gilliver wrote:
    fails, the result can much severe than the degradation of an analogue
    signal.
    Oh yes, the "digital cliff", where the error correction fails. Although
    the improvement in going to (corrected!) digital is vast (and gives me
    many more channels), I find it_much_ more irritating when it_does_
    fail - for video - than a noisy analogue channel. I think it's because
    there is so much less warning, so it's so sudden.

    OTOH digital with error correct maintains quality well beyond what an
    analogue signal can.

    UIf you are getting degradation the signal level is already so corrupt
    or low that no analogue is likely to be useable
    --
    The lifetime of any political organisation is about three years before
    its been subverted by the people it tried to warn you about.

    Anon.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Thu Apr 2 14:16:04 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 10:43, Richmond wrote:
    I found this:

    https://www.vodafone.co.uk/newscentre/press-release/vodafone-brings-hd-voice-call-quality-to-customers/

    "Vodafone UK has rolled out High Definition Voice technology across the UK Customers making Vodafone-to-Vodafone calls on compatible phones will
    enjoy exceptionally clear conversations with family, friends and work colleagues"

    Unfortunately their customer service is intolerable to me.

    Yeah. With some companies your only recourse when problems exist is to
    simply cease to use them.
    --
    The lifetime of any political organisation is about three years before
    its been subverted by the people it tried to warn you about.

    Anon.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Thu Apr 2 23:04:29 2026
    From Newsgroup: uk.telecom

    On 2026/4/2 14:0:24, The Natural Philosopher wrote:
    On 02/04/2026 01:08, J. P. Gilliver wrote:
    On 2026/4/1 23:54:29, David Woolley wrote:
    []
    Regarding another reply, "codec" is often used to mean the output format >>> produced by a codec, rather than the the actual process that does the
    coding and decoding, but the word "codec" is a combination of code and
    decode, so refers to the process.

    I thought it was a combination of coder and decoder, i. e. it's the
    actual hardware (or software, or firmware).

    Of course it is. Certain people don't know as much as they want you to believe they do,

    Codec is a device that interfaces between generally analogue signals of
    some sort at one end and a digital data stream at the other. How that digital data stream is transferred is implementation specific. As is
    any signal processing, digital or analogue, that is carried out by the aforesaid codec.

    It's mostly applied to software layers these days with the actual
    digital to analogue (and vice versa) being performed by devices with
    other names, like 'modem' or 'DAC'

    Strictly, I don't think of the ADC/DAC as part of the CoDec; on the
    whole, I wouldn't think of analogue processing stages as part of it
    either. but I can see that they could be. Academic nowadays - and for
    some time - anyway; compression nearly always done by software
    (occasionally firmware), in the digital domain, rather than a piece of hardware.

    Never thought of MoDem as part of it eiter, but I suppose it could be.



    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    If you are afraid of being lonely, don't try to be right.
    - Jules Renard, writer (1864-1910)
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Thu Apr 2 23:21:18 2026
    From Newsgroup: uk.telecom

    On 2026/4/2 13:53:48, The Natural Philosopher wrote:
    []
    All comms is a (digital) signal imposed on an analogue carrier. Unless

    A lot of modern schemes, I'd say "on several analogue carriers". Sure,
    you can insist it's one, but that makes the nature of the digital
    signal, and the imposing, considerably more complex.

    you are counting atoms and electrons individually *everything* is
    analogue at the bottom.

    :-)
    []
    In general the old baseband phones were converted to 64kbps synchronous
    TDM with iirc an 8 bit 8kHz sample rate giving an effective upper
    frequency limit of about 3Khz and S/N ratio of about 45dB. I*n those
    days (1960s?)the computer power to do compression didn't exist and nor
    did the algorithms

    did it even do error-correction?

    It's very easy to do a lot better than that with a compressed VOIP
    signal, and in today's broadband world you can go a LOT faster than
    64kbps if you want. And voice, with distinct lack of much at the top
    end of the frequency band and plenty of pauses, and only a single
    frequency set, is absolutely ideal for compression.

    I heard once (apocryphal) that intelligible speech could be transmitted
    at 50 bytes per second.

    I could believe it, though I imagine grotty, though intelligible. (you
    might not be able to identify the speaker, for example.)

    G722 is actually slightly above the old 8kHz sample and the bandwidth extends to 7kHz. And does use compression because the final bitrate is

    7 kHz implies 14+kHz sample rate (unless throwing away some of the
    bottom and mixing).

    variable at 16, 24 and 32 kbps

    I have no idea if it is the 'general standard' but one things is clear,
    that excluding *mobile* (RF) phone technology, VOIP over broadband is superior to a baseband phone in every respect.



    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    If you are afraid of being lonely, don't try to be right.
    - Jules Renard, writer (1864-1910)
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Thu Apr 2 23:24:42 2026
    From Newsgroup: uk.telecom

    On 2026/4/2 14:14:22, The Natural Philosopher wrote:
    On 02/04/2026 01:14, J. P. Gilliver wrote:
    fails, the result can much severe than the degradation of an analogue
    signal.
    Oh yes, the "digital cliff", where the error correction fails. Although
    the improvement in going to (corrected!) digital is vast (and gives me
    many more channels), I find it_much_ more irritating when it_does_
    fail - for video - than a noisy analogue channel. I think it's because
    there is so much less warning, so it's so sudden.

    OTOH digital with error correct maintains quality well beyond what an analogue signal can.

    UIf you are getting degradation the signal level is already so corrupt
    or low that no analogue is likely to be useable


    True. It's just that error correction hides the _fact_ of the
    degradation - until it "falls off the cliff" _suddenly_. With analogue,
    you get some _warning_ that something is going wrong (and, occasionally,
    can do something about it, such as moving the aerial, or attending to
    the source of the interference); with (corrected) digital, you have
    little or no hint until it's too late.
    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    If you are afraid of being lonely, don't try to be right.
    - Jules Renard, writer (1864-1910)
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Apr 3 12:17:06 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 23:04, J. P. Gilliver wrote:
    Strictly, I don't think of the ADC/DAC as part of the CoDec; on the
    whole, I wouldn't think of analogue processing stages as part of it
    either. but I can see that they could be. Academic nowadays - and for
    some time - anyway; compression nearly always done by software
    (occasionally firmware), in the digital domain, rather than a piece of hardware.

    Never thought of MoDem as part of it eiter, but I suppose it could be.

    When you learn network theory there is somewhere a 7 layer model.

    When you end up designing real kit all that goes out the window, The
    layers are blurred, part of 'layer 2' is being done by the hardware
    that 'does layer 3'; etc etc.

    Aeons ago when I first joined a company as an apprentice to learn
    electronics as it was practiced, I was given some boards to service,
    from a radar rack.

    I hazed in amazement at a board with two capacitors, thirteen resistors
    and one op amp chip covering an area of around 46 square inches with a
    gold plated edge connector that cost more than the rest of the board
    put together...

    I finally figured it out., A boffin had drawn a *block diagram* of the
    whole radar set and one of the blocks was an analogue adder.

    Then the job had been passed to a technician who had assigned each
    block, a circuit board, whether it needed it or not, because he was
    thick and lazy. And money was no object.

    And another technician had dutifully taken that block and turned it into
    a circuit board.
    And that is how the world works.

    This isn't about technology - this codec thing, Its about word
    definition, Semantics. One builds circuits, they do stuff. They have
    hardware and software. Somehow they get fiddled with till they work.

    Then people argue over what to call them - where to draw the line
    between 'codec' and 'DAC' , and whether a filter in hardware that does
    the same thing as one in software is in fact the same thing.

    Because stupid people want simple stories, that make them look smart. So
    they argue over definitions and pretend it's about technology.

    TLDR. VOIP is coming. Over broadband. You dont need to know how it
    works, just that it does, and how to use it, and to know that in general
    its better quality than old school TDM telephony. Because it uses
    better tech and more of it.

    The specifications of the interchange protocols like G722 are what they
    are, and designed by people with better math skills than any of us ,
    including me.

    Whether the circuits that perform the translation between G722 and
    baseband audio are (called) codecs, DACs ,ADCs digital signal processors
    or orgasmatrons, is supremely irrelevant unless you want to try to
    appear smarter than you are.

    Just let it drop.
    --
    A lie can travel halfway around the world while the truth is putting on
    its shoes.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Apr 3 12:24:30 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 23:21, J. P. Gilliver wrote:
    On 2026/4/2 13:53:48, The Natural Philosopher wrote:
    []
    All comms is a (digital) signal imposed on an analogue carrier. Unless

    A lot of modern schemes, I'd say "on several analogue carriers". Sure,
    you can insist it's one, but that makes the nature of the digital
    signal, and the imposing, considerably more complex.

    you are counting atoms and electrons individually *everything* is
    analogue at the bottom.

    :-)
    []
    In general the old baseband phones were converted to 64kbps synchronous
    TDM with iirc an 8 bit 8kHz sample rate giving an effective upper
    frequency limit of about 3Khz and S/N ratio of about 45dB. I*n those
    days (1960s?)the computer power to do compression didn't exist and nor
    did the algorithms

    did it even do error-correction?
    Well that is an intersting point. Certainnly at the level of conversion
    to a 64kbps data steam it did not
    BUT those 64k data streams got consolidated into T1 and E1 streams and
    it is likely that later on peoplee ran error correction on those

    Its been a few decades...



    It's very easy to do a lot better than that with a compressed VOIP
    signal, and in today's broadband world you can go a LOT faster than
    64kbps if you want. And voice, with distinct lack of much at the top
    end of the frequency band and plenty of pauses, and only a single
    frequency set, is absolutely ideal for compression.

    I heard once (apocryphal) that intelligible speech could be transmitted
    at 50 bytes per second.

    I could believe it, though I imagine grotty, though intelligible. (you
    might not be able to identify the speaker, for example.)
    Oh absolutely, Its more like resynthesising speech from tokens
    (phonemes?) like an AI does,


    G722 is actually slightly above the old 8kHz sample and the bandwidth
    extends to 7kHz. And does use compression because the final bitrate is

    7 kHz implies 14+kHz sample rate (unless throwing away some of the
    bottom and mixing).

    Actually in practical terms it implies about 16kHz and I am sure that's
    what they do..its a nice round digital sounding sample frequency :-)

    My guess is that it is more than 8 bits at the raw sample as well, and
    the compression algo. is chosen to match the energy spectra of speech
    and music,


    variable at 16, 24 and 32 kbps

    I have no idea if it is the 'general standard' but one things is clear,
    that excluding *mobile* (RF) phone technology, VOIP over broadband is
    superior to a baseband phone in every respect.



    --
    The New Left are the people they warned you about.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Fri Apr 3 12:26:14 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 23:24, J. P. Gilliver wrote:
    True. It's just that error correction hides the_fact_ of the
    degradation - until it "falls off the cliff"_suddenly_. With analogue,
    you get some_warning_ that something is going wrong (and, occasionally,
    can do something about it, such as moving the aerial, or attending to
    the source of the interference); with (corrected) digital, you have
    little or no hint until it's too late.

    Well if you care enough you can build in a 'quality;' monitor on the
    incoming signal...
    Raw error bit rate is a typical metric. How many times did you have to
    error correct?
    --
    "An intellectual is a person knowledgeable in one field who speaks out
    only in others...rCY

    Tom Wolfe

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Fri Apr 3 23:03:20 2026
    From Newsgroup: uk.telecom

    On 2026/4/3 12:17:6, The Natural Philosopher wrote:
    []
    When you learn network theory there is somewhere a 7 layer model.

    Oh yes, I think I came across that.

    When you end up designing real kit all that goes out the window, The
    layers are blurred, part of 'layer 2' is being done by the hardware
    that 'does layer 3'; etc etc.

    Yes, I tended to think the "layers" somewhat arbitrary when you come to
    the actual hardware, though I suppose _if you had time_ they made the
    design documents neater.

    Aeons ago when I first joined a company as an apprentice to learn electronics as it was practiced, I was given some boards to service,
    from a radar rack.

    I hazed in amazement at a board with two capacitors, thirteen resistors
    and one op amp chip covering an area of around 46 square inches with a
    gold plated edge connector that cost more than the rest of the board
    put together...

    I finally figured it out., A boffin had drawn a *block diagram* of the
    whole radar set and one of the blocks was an analogue adder.

    Then the job had been passed to a technician who had assigned each
    block, a circuit board, whether it needed it or not, because he was
    thick and lazy. And money was no object.

    :-)

    And another technician had dutifully taken that block and turned it into
    a circuit board.
    And that is how the world works.

    This isn't about technology - this codec thing, Its about word
    definition, Semantics. One builds circuits, they do stuff. They have hardware and software. Somehow they get fiddled with till they work.

    Indeed.

    Then people argue over what to call them - where to draw the line
    between 'codec' and 'DAC' , and whether a filter in hardware that does
    the same thing as one in software is in fact the same thing.

    I was at one point quite involved in DSP - initially dealing with HF communications where the multipath variation was several symbols long
    (you deal with that by either channel esttmation or channel
    equalization; there are path conditions that kill both [we worked on estimation, IIRR]); later with vibration cancellation for marine
    "machinery rafts". All good fun.

    Because stupid people want simple stories, that make them look smart. So they argue over definitions and pretend it's about technology.

    We were more involved with getting the things to work (both hardware and software, though of course latterly more the latter).

    TLDR. VOIP is coming. Over broadband. You dont need to know how it
    works, just that it does, and how to use it, and to know that in general
    its better quality than old school TDM telephony. Because it uses
    better tech and more of it.

    The specifications of the interchange protocols like G722 are what they
    are, and designed by people with better math skills than any of us , including me.

    I'm pretty sure it will involve buying an ATA in my case, thus involving
    the invoking of an SEP field.

    Whether the circuits that perform the translation between G722 and
    baseband audio are (called) codecs, DACs ,ADCs digital signal processors
    or orgasmatrons, is supremely irrelevant unless you want to try to
    appear smarter than you are.

    Just let it drop.

    I certainly wasn't trying to preserve those definitions; I've long since
    given up trying to fix the errors propagated (including by the industry)
    by both the general public and those who should know better. (Such as
    calling the box that contains everything a "router", at least until FTTP started to become common; that one _helps_ the general public.)


    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    As individuals, politicians are usually quite charming, so it is quite
    hard to dislike them, but in most cases, it is worth making the effort.
    - Mark Williams (UMRA), 2013-4-26
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From J. P. Gilliver@G6JPG@255soft.uk to uk.telecom on Fri Apr 3 23:12:22 2026
    From Newsgroup: uk.telecom

    On 2026/4/3 12:26:14, The Natural Philosopher wrote:
    On 02/04/2026 23:24, J. P. Gilliver wrote:
    True. It's just that error correction hides the_fact_ of the
    degradation - until it "falls off the cliff"_suddenly_. With analogue,
    you get some_warning_ that something is going wrong (and, occasionally,
    can do something about it, such as moving the aerial, or attending to
    the source of the interference); with (corrected) digital, you have
    little or no hint until it's too late.

    Well if you care enough you can build in a 'quality;' monitor on the incoming signal...
    Raw error bit rate is a typical metric. How many times did you have to
    error correct?


    I remember in the very early days of CD players, there were some -
    really only lab models; I was going to say expensive ones, but in those
    days they were _all_ expensive - that had a digital quality, or error
    rate, indication (possibly just a light).

    I remember also in the work I did, we had something to insert known data
    into the channel - we called it Bit Error Rate Tester, or BERT for short.
    --
    J. P. Gilliver. UMRA: 1960/<1985 MB++G()ALIS-Ch++(p)Ar++T+H+Sh0!:`)DNAf

    As individuals, politicians are usually quite charming, so it is quite
    hard to dislike them, but in most cases, it is worth making the effort.
    - Mark Williams (UMRA), 2013-4-26
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Sat Apr 4 19:12:30 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 10:13, Richmond wrote:
    So really the question is, with the right SIP/VOIP telephone equipment,
    and a VOIP to VOIP connection, can I achieve this full frequency?

    There is no point in having the low frequencies; they don't contribute
    to voice intelligibility (and, incidentally, in an amateur radio
    context, they are undesirable, because they consume a lot of power, in analogue systems, in a context where the licence limits the total power).

    The problem that G.722 addresses, is transmitting the higher speech frequencies, which allow one to distinguish between s, sh, and h, etc.,
    but aren't weren't carried on analogue carrier systems, nor are they
    carried on the 8kHz sampling channels that was used in the original
    digital phone systems. It's aim is to do this in the same digital
    bandwidth as was already in use, and to not be a speech only codec, and
    only adding an acceptable amount of extra quantisation noise.

    It is the higher end where the analogue phone system had problems. AM
    radio is slightly better. FM broadcast radio is quite good, covering frequencies only really useful for music.

    The compression G.722 uses is ADPCM (Adaptive Differential Pulse Code Modulation).

    That's compression relative to 16 bit linear PCM (probably less than 16
    bits, as I don't think it can actually make use of all of them), but the digital part of the analogue phone network, also compressed with respect
    to linear PCM, in that case by using somewhat logarithmic transfer
    functions, which is what G.711 is.

    I do
    have FTTP

    G.722 uses exactly the same digital bandwidth as G.711, so it will work comfortably on borderline ADSL. I'm not aware of any audio codec that
    needs more than ADSL bit rates. I suppose full CD quality 16bit linear, 44.1kHz, might be iffy in the uplink direction, but, except possibly for broadcasters, no-one would use that.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Sat Apr 4 19:21:32 2026
    From Newsgroup: uk.telecom

    On 02/04/2026 01:14, J. P. Gilliver wrote:
    In that "digital signals" always, for most of about the last
    fortysomething years, means "digital signals with error correction";
    this is so often assumed as to be understood, but the omission of the
    fact confuses newcomers to the subject (who are left wondering_why_ are digital signals better - and/or more reliable - until this point is explained. Which it often isn't).

    Error correction allows you to get closer to the Shannon Hartley limit
    for Gaussian noise, and can also cope with short dropouts, but I think
    the key original benefit was that it could be regenerated almost
    perfectly, whereas analogue would always accumulate noise and distortion.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Sun Apr 5 00:08:45 2026
    From Newsgroup: uk.telecom

    David Woolley <david@ex.djwhome.demon.invalid> writes:

    On 02/04/2026 10:13, Richmond wrote:
    So really the question is, with the right SIP/VOIP telephone equipment,
    and a VOIP to VOIP connection, can I achieve this full frequency?

    There is no point in having the low frequencies; they don't contribute
    to voice intelligibility

    There is no point if all you want is to understand the speech, but it
    does change the sound and make it more real and immediate, and I
    appreciate that, I think it is worth doing, especially as we now have
    band width coming out of our ears with fibre.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Sun Apr 5 12:01:06 2026
    From Newsgroup: uk.telecom

    On 04/04/2026 19:12, David Woolley wrote:
    The problem that G.722 addresses, is transmitting the higher speech frequencies, which allow one to distinguish between s, sh, and h, etc.,
    but aren't weren't carried on analogue carrier systems, nor are they
    carried on the 8kHz sampling channels that was used in the original
    digital phone systems.

    Whilst sibilants in >4KHz range carry a lot of *fidelity* they do not in
    any sense contribute to intelligibility.
    --
    rCLThere are two ways to be fooled. One is to believe what isnrCOt true; the other is to refuse to believe what is true.rCY

    rCoSoren Kierkegaard

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Sun Apr 5 12:10:15 2026
    From Newsgroup: uk.telecom

    On 05/04/2026 00:08, Richmond wrote:
    David Woolley <david@ex.djwhome.demon.invalid> writes:

    On 02/04/2026 10:13, Richmond wrote:
    So really the question is, with the right SIP/VOIP telephone equipment,
    and a VOIP to VOIP connection, can I achieve this full frequency?

    There is no point in having the low frequencies; they don't contribute
    to voice intelligibility

    There is no point if all you want is to understand the speech, but it
    does change the sound and make it more real and immediate, and I
    appreciate that, I think it is worth doing, especially as we now have
    band width coming out of our ears with fibre.

    Low frequencies take up no bandwidth at all, really.
    So many people talking through their backsides here.

    Telephones have always been able to carry low frequencies but the old
    carbon mics couldn't generate them and nor could the primitive earpieces reproduce them.

    A long tome ago simple tests revealed that provided you could send
    300Hz-3kHz adequately, speech was intelligible and clear.

    People who have actually pursued audio engineering fir a living know
    that above 3kHz is just the hisses and clicks of the sound, and the fundamentals of voice do not go below 200Hz, That range is reserved for
    bass (guitars) and drums, organs and the odd large woodwind.

    Cassette recorders were very hard to get above 5kHz and many people
    still think they were 'hifi'
    --
    All political activity makes complete sense once the proposition that
    all government is basically a self-legalising protection racket, is
    fully understood.


    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Chris Green@cl@isbd.net to uk.telecom on Sun Apr 5 12:53:47 2026
    From Newsgroup: uk.telecom

    The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 04/04/2026 19:12, David Woolley wrote:
    The problem that G.722 addresses, is transmitting the higher speech frequencies, which allow one to distinguish between s, sh, and h, etc., but aren't weren't carried on analogue carrier systems, nor are they carried on the 8kHz sampling channels that was used in the original digital phone systems.

    Whilst sibilants in >4KHz range carry a lot of *fidelity* they do not in
    any sense contribute to intelligibility.

    Well that's not true in my experience, I've lost quite a lot of
    hearing at the high frequency end and that does make it difficult to differentiate between, say, 'sallow' and 'fallow', or 'cyst' and
    'fist'. (there's probably better examples but I can't think of them
    now). Obviously context helps a lot but when my wife suggests an
    answer to a crossword clue there isn't any context and that's when I
    have difficulty.
    --
    Chris Green
    -+
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Sun Apr 5 13:22:21 2026
    From Newsgroup: uk.telecom

    On 05/04/2026 12:53, Chris Green wrote:
    The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 04/04/2026 19:12, David Woolley wrote:
    The problem that G.722 addresses, is transmitting the higher speech
    frequencies, which allow one to distinguish between s, sh, and h, etc.,
    but aren't weren't carried on analogue carrier systems, nor are they
    carried on the 8kHz sampling channels that was used in the original
    digital phone systems.

    Whilst sibilants in >4KHz range carry a lot of *fidelity* they do not in
    any sense contribute to intelligibility.

    Well that's not true in my experience, I've lost quite a lot of
    hearing at the high frequency end

    People with hearing loss lose it above 800Hz. check your charts. Also
    they lose the frigging LOT to a large extent.

    and that does make it difficult to
    differentiate between, say, 'sallow' and 'fallow', or 'cyst' and
    'fist'. (there's probably better examples but I can't think of them
    now). Obviously context helps a lot but when my wife suggests an
    answer to a crossword clue there isn't any context and that's when I
    have difficulty.

    As I said, you probably don't have much above 1kHz, You do need to go to
    3kHz+ if you can.

    My Ex's hearing aid started at 600 and peaked at 2kHz. She couldn't hear
    a damn thing above that.


    Old AM (MW. LW) radio was very very lucky to get to anywhere near 4kHz,
    at least in Europe where 9kHz was the channel spacing. And yet it was
    pretty intelligible to most of us.

    Old age brings some or all of
    - reduction in total sensitivity of up to 60dB or more
    - loss of higher frequencies above say 2kHz making some sibilants hard
    to pick out
    - intermodulation distortion making it hard to pick out one voice from
    another in a crowd.
    --
    "Socialist governments traditionally do make a financial mess. They
    always run out of other people's money. It's quite a characteristic of them"

    Margaret Thatcher

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From liz@liz@poppyrecords.invalid.invalid (Liz Tuddenham) to uk.telecom on Sun Apr 5 14:48:53 2026
    From Newsgroup: uk.telecom

    The Natural Philosopher <tnp@invalid.invalid> wrote:

    [...]
    Telephones have always been able to carry low frequencies but the old
    carbon mics couldn't generate them and nor could the primitive earpieces reproduce them.

    Telephone lines can carry down to DC but the coupling capacitors in the telephone exchange, and the inductance of the relay coils that delivered
    power to the line, were chosen on the basis of a 300 c/s cutoff.

    Actually some carbon mics give a good response at frequencies well below
    300 c/s but those in telephone mouthpieces were designed not to.
    Telephone earpieces of the diaphragm type could cope with low
    frequencies (if you could get an airtight seal between the bakelite
    cover and your ear) but they had a lot of resonanaces which gave them a characteristic sound.


    A long tome ago simple tests revealed that provided you could send 300Hz-3kHz adequately, speech was intelligible and clear.

    That seems to have been forgotten in modern 'phones, some of which
    barely extend above 1 Kc/s. It all makes money for the companies that
    charge by call time:

    "I've had a problem with the grmbmfgg."
    "Sorry, what did you say?"
    " i said I've had a problem."
    "Yes, I heard that but but what was it with ?"
    "What about a pig???"
    "No, what was it WITH?"
    "What was what with?"
    "The problem..."
    "Yes, I've had one"

    etc.....
    --
    ~ Liz Tuddenham ~
    (Remove the ".invalid"s and add ".co.uk" to reply)
    www.poppyrecords.co.uk
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From David Woolley@david@ex.djwhome.demon.invalid to uk.telecom on Sun Apr 5 18:07:05 2026
    From Newsgroup: uk.telecom

    On 05/04/2026 12:01, The Natural Philosopher wrote:
    Whilst sibilants in >4KHz range carry a lot of *fidelity* they do not in
    any sense contribute to intelligibility.

    That's not the position normally taken on the "speech banana" model <https://en.wikipedia.org/wiki/Speech_banana>. Sibilants are above
    4kHz, so the 3.4kHz cut-off for normal telephone audio requires the
    brain to guess from context.

    I think even low end hearing aids are designed with a 6kHz cut- off,
    these days.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From The Natural Philosopher@tnp@invalid.invalid to uk.telecom on Sun Apr 5 21:14:29 2026
    From Newsgroup: uk.telecom

    On 05/04/2026 18:07, David Woolley wrote:
    On 05/04/2026 12:01, The Natural Philosopher wrote:
    Whilst sibilants in >4KHz range carry a lot of *fidelity* they do not
    in any sense contribute to intelligibility.

    That's not the position normally taken on the "speech banana" model <https://en.wikipedia.org/wiki/Speech_banana>.-a Sibilants are above
    4kHz, so the 3.4kHz cut-off for normal telephone audio requires the
    brain to guess from context.

    I am talking from experience., not recieved wisdom

    I think even low end hearing aids are designed with a 6kHz cut--a off,
    these days.

    If you need a hearing aid at all the chances are you have fuck all above
    2kHz left

    I remember my exe'es dad getting his and complaining he could 'now hear
    his trousers rubbing together as he walked' But he couldn't hear the
    clock ticking, even so..

    I think many people become deaf in order not to listen to other people.
    His wife would definitely have been on my murder list if she had been
    mine...
    --
    The New Left are the people they warned you about.

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Tue Mar 17 22:45:38 2026
    From Newsgroup: uk.telecom

    I was investigating whether using a proper SIP phone, as opposed to a
    DECT phone plugged into an ATA, would provide a better sound quality,
    including the lower pitch sounds. It seems most VOIP providers don't
    support G.722 which is the compression algorithm which would provide
    this. Perhaps this is because in most cases calls would be going to
    analog lines. But now it seems that 78% of homes have access to full
    fibre. Does this mean they are actually using it though? Well apparently
    not, the actual uptake is less than half, maybe 39%.

    But PSTN is only being used by about a quarter of landlines. (This
    figure is hard to be certain about because VM were using some other
    analog system EMTA.)

    So why not switch on G.722?

    https://www.ofcom.org.uk/phones-and-broadband/telecoms-infrastructure/new-ofcoms-final-push-towards-full-fibre-finish-line-to-help-bolster-uk-productivity
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Marco Moock@mm@dorfdsl.de to uk.telecom on Wed Mar 18 20:42:00 2026
    From Newsgroup: uk.telecom

    On 17.03.2026 22:45 Uhr Richmond wrote:

    I was investigating whether using a proper SIP phone, as opposed to a
    DECT phone plugged into an ATA, would provide a better sound quality, including the lower pitch sounds. It seems most VOIP providers don't
    support G.722 which is the compression algorithm which would provide
    this. Perhaps this is because in most cases calls would be going to
    analog lines. But now it seems that 78% of homes have access to full
    fibre. Does this mean they are actually using it though? Well
    apparently not, the actual uptake is less than half, maybe 39%.

    But PSTN is only being used by about a quarter of landlines. (This
    figure is hard to be certain about because VM were using some other
    analog system EMTA.)

    So why not switch on G.722?

    I can only speak for Germany, but I assume the situation won't much
    differ in the UK.

    Most people use analog phones connected to their ATA RJ11 jack at the
    router, so there won't be better audio quality for them.

    SIP phones for home users are a corner cases and ISDN phones were also
    phased out, many current CPEs vendors don't provide S0 anymore or never
    did.

    G.711 works in most cases with most equipment. If all use that codec,
    there is no need for a translation between the ISPs.

    I assume even if G.722 were used by default, the amount of phone calls
    where that codec could be used (both ISPs support it, including the
    transit interconnection ISPs, both customer CPEs have it enabled, both customers use SIP/DECT/ISDN phones that have proper microphones), would
    be low.
    --
    kind regards
    Marco

    Send spam to 1773783938muell@stinkedores.dorfdsl.de

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Wed Mar 18 21:32:41 2026
    From Newsgroup: uk.telecom

    Marco Moock <mm@dorfdsl.de> writes:

    On 17.03.2026 22:45 Uhr Richmond wrote:

    I was investigating whether using a proper SIP phone, as opposed to a
    DECT phone plugged into an ATA, would provide a better sound quality,
    including the lower pitch sounds. It seems most VOIP providers don't
    support G.722 which is the compression algorithm which would provide
    this. Perhaps this is because in most cases calls would be going to
    analog lines. But now it seems that 78% of homes have access to full
    fibre. Does this mean they are actually using it though? Well
    apparently not, the actual uptake is less than half, maybe 39%.

    But PSTN is only being used by about a quarter of landlines. (This
    figure is hard to be certain about because VM were using some other
    analog system EMTA.)

    So why not switch on G.722?

    I can only speak for Germany, but I assume the situation won't much
    differ in the UK.

    Most people use analog phones connected to their ATA RJ11 jack at the
    router, so there won't be better audio quality for them.

    SIP phones for home users are a corner cases and ISDN phones were also
    phased out, many current CPEs vendors don't provide S0 anymore or
    never did.

    G.711 works in most cases with most equipment. If all use that codec,
    there is no need for a translation between the ISPs.

    I assume even if G.722 were used by default, the amount of phone calls
    where that codec could be used (both ISPs support it, including the
    transit interconnection ISPs, both customer CPEs have it enabled, both customers use SIP/DECT/ISDN phones that have proper microphones),
    would be low.

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol) doesn't necessarily choose the highest level, because it considers quality and
    latency. Progress seems slow. Signal has a better frequency range but
    has the usual drop outs.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Marco Moock@mm@dorfdsl.de to uk.telecom on Thu Mar 19 19:56:52 2026
    From Newsgroup: uk.telecom

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol) doesn't necessarily choose the highest level, because it considers quality and latency. Progress seems slow. Signal has a better frequency range but
    has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to avoid
    for hundreds of calls, especially if the other side doesn't have any
    advantage.
    --
    kind regards
    Marco

    Send spam to 1773865961muell@stinkedores.dorfdsl.de

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Fri Mar 20 10:48:39 2026
    From Newsgroup: uk.telecom

    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it considers
    quality and latency. Progress seems slow. Signal has a better
    frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't have
    any advantage.

    Why does the ISP need to do processing?

    This sort of thing seems insignificant to me, compared to streaming
    audio and video.

    The problems I think are NAT, Firewalls, and publicity. If you know
    someone's SIP address you ought to be able to make a direct call over
    the internet in the same way as Signal does. It would be free of any per
    minute cost. The compression would be done with equipment at either end,
    not by the ISP. That's how it looks to me anyway.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Fri Mar 20 11:23:46 2026
    From Newsgroup: uk.telecom

    Richmond <dnomhcir@gmx.com> writes:

    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it considers
    quality and latency. Progress seems slow. Signal has a better
    frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't have
    any advantage.

    Why does the ISP need to do processing?

    This sort of thing seems insignificant to me, compared to streaming
    audio and video.

    The problems I think are NAT, Firewalls, and publicity. If you know
    someone's SIP address you ought to be able to make a direct call over
    the internet in the same way as Signal does. It would be free of any per minute cost. The compression would be done with equipment at either end,
    not by the ISP. That's how it looks to me anyway.

    https://nickvsnetworking.com/the-sad-story-of-enum-in-australia/

    "ENUM was going to change telephone routing. No longer would you need to
    pay a carrier to take your calls across the PSTN, but rather through the
    use of DNS your handset would look up a destination and route in a peer
    to peer fashion.

    Number porting would just be a matter of updating NAPTR records, almost
    all calls would be free as thererCOs no way/need to charge and media would
    flow directly from the calling party to the called party."

    But instead people are driven into the hands of Meta monopoly.
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Andy Burns@usenet@andyburns.uk to uk.telecom on Fri Mar 20 11:58:39 2026
    From Newsgroup: uk.telecom

    Richmond wrote:

    If you know someone's SIP address you ought to be able to make a
    direct call over the internet in the same way as Signal does. It
    would be free of any per minute cost. The compression would be done
    with equipment at either end, not by the ISP.

    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather than
    via the provider.

    But I don't think you'd want a tidal wave of AI vishing calls would you?
    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Marco Moock@mm@dorfdsl.de to uk.telecom on Fri Mar 20 16:01:28 2026
    From Newsgroup: uk.telecom

    On 20.03.2026 10:48 Uhr Richmond wrote:

    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it considers
    quality and latency. Progress seems slow. Signal has a better
    frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't
    have any advantage.

    Why does the ISP need to do processing?

    If the 2 subscribers use different codes, the signal needs to be
    converted.

    This sort of thing seems insignificant to me, compared to streaming
    audio and video.

    But it needs CPU power for processing.

    The problems I think are NAT, Firewalls, and publicity.

    Indeed they are. That's why many SIP services support IPv6 or run
    directly on the CPE that is being assigned a public IPv4.

    If you know someone's SIP address you ought to be able to make a
    direct call over the internet in the same way as Signal does. It
    would be free of any per minute cost. The compression would be done
    with equipment at either end, not by the ISP. That's how it looks to
    me anyway.

    This is possible, but uncommon. There are concepts to handle that via
    DNS, but that means the telcos cannot make money (because you can use
    any SIP (soft)phone for outgoing calls), so they chose not to use that.
    --
    kind regards
    Marco

    Send spam to 1774000119muell@stinkedores.dorfdsl.de

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Marco Moock@mm@dorfdsl.de to uk.telecom on Fri Mar 20 16:03:08 2026
    From Newsgroup: uk.telecom

    On 20.03.2026 11:58 Uhr Andy Burns wrote:

    The call setup (SIP) is done via the VoIP provider, but normally the
    actual speech (RTP) is direct between the caller and callee, rather
    than via the provider.

    I haven't seen that. Are there SIP setups that handle it this way?

    I have set up my SIP service using IPv6 and have not opened ports on
    IPv4, but I can do phone calls to people with only IPv4, so there
    cannot be a direct connection in that case.
    --
    kind regards
    Marco

    Send spam to 1774004319muell@stinkedores.dorfdsl.de

    --- Synchronet 3.21f-Linux NewsLink 1.2
  • From Richmond@dnomhcir@gmx.com to uk.telecom on Fri Mar 20 15:22:11 2026
    From Newsgroup: uk.telecom

    Marco Moock <mm@dorfdsl.de> writes:

    On 20.03.2026 10:48 Uhr Richmond wrote:

    Marco Moock <mm@dorfdsl.de> writes:

    On 18.03.2026 21:32 Uhr Richmond wrote:

    I thought that the highest compression would be negotiated
    automatically. But it seems SDP (Session Description Protocol)
    doesn't necessarily choose the highest level, because it considers
    quality and latency. Progress seems slow. Signal has a better
    frequency range but has the usual drop outs.

    Conversation means CPU power is needed, something the ISP want to
    avoid for hundreds of calls, especially if the other side doesn't
    have any advantage.

    Why does the ISP need to do processing?

    If the 2 subscribers use different codes, the signal needs to be
    converted.

    But they can negotiate the same codes.
    --- Synchronet 3.21f-Linux NewsLink 1.2