From Newsgroup: rec.audio.pro

Not sure if anyone still monitors this group, but I have a question if so.

I have been using some rather extraordinary online models and algorithms
to denoise some wildlife audio recordings I made several years ago. The
most startling thing I have found is how WELL these models do. About
half of my recordings I considered unusable. Low in the noise and if
software NR was applied, there were too many artifacts. No longer!

So, as I'm not used to having completely noiseless results, I wonder how
much noise I should mix back in? Two channels are created when using
the online denoisers: the denoised track (and I mean denoised!!) and
just the noise it removed.

I come from an age of cassette tape which, even with Dolby NR, there was always noise. Even when I used my software NR, I always mixed some back
in to help hide artifacts.

Just some general suggestions welcome. Thanks in advance.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: rec.audio.pro

On 9/13/25 1:04 AM, Harvey Sanenbum wrote:

Not sure if anyone still monitors this group, but I have a question if so.

I have been using some rather extraordinary online models and algorithms
to denoise some wildlife audio recordings I made several years ago.-a The most startling thing I have found is how WELL these models do.-a About
half of my recordings I considered unusable.-a Low in the noise and if software NR was applied, there were too many artifacts.-a No longer!

So, as I'm not used to having completely noiseless results, I wonder how much noise I should mix back in?-a Two channels are created when using
the online denoisers: the denoised track (and I mean denoised!!) and
just the noise it removed.

I come from an age of cassette tape which, even with Dolby NR, there was always noise.-a Even when I used my software NR, I always mixed some back
in to help hide artifacts.

Just some general suggestions welcome.-a Thanks in advance.

Just wanted to note that, according to the meter, I mixed about -52 dB
of noise back into the "noiseless" result.
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: rec.audio.pro

On 13/09/2025 5:09 pm, Harvey Sanenbum wrote:

On 9/13/25 1:04 AM, Harvey Sanenbum wrote:

Not sure if anyone still monitors this group, but I have a question if
so.

I have been using some rather extraordinary online models and
algorithms to denoise some wildlife audio recordings I made several
years ago.-a The most startling thing I have found is how WELL these
models do.-a About half of my recordings I considered unusable.-a Low in
the noise and if software NR was applied, there were too many
artifacts.-a No longer!

So, as I'm not used to having completely noiseless results, I wonder
how much noise I should mix back in?-a Two channels are created when
using the online denoisers: the denoised track (and I mean denoised!!)
and just the noise it removed.

I come from an age of cassette tape which, even with Dolby NR, there
was always noise.-a Even when I used my software NR, I always mixed
some back in to help hide artifacts.

Just some general suggestions welcome.-a Thanks in advance.

Just wanted to note that, according to the meter, I mixed about -52 dB
of noise back into the "noiseless" result.

Sounds more natural ?
--
geoff
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: rec.audio.pro

Geoff <geoff@nospamgeoffwood.org> writes, quoting Harvey Sanenbum <harvey50120@micro.net>:

I have been using some rather extraordinary online models and
algorithms to denoise some wildlife audio recordings I made several
years ago.? The most startling thing I have found is how WELL thes
models do.? About half of my recordings I considered unusable.? L
in the noise and if software NR was applied, there were too many
artifacts.? No longer

So, as I'm not used to having completely noiseless results, I wonder
how much noise I should mix back in?? Two channels are created whe
using the online denoisers: the denoised track (and I mean denoised!!)
and just the noise it removed.

I come from an age of cassette tape which, even with Dolby NR, there
was always noise.? Even when I used my software NR, I always mixe
some back in to help hide artifacts.

Just some general suggestions welcome.? Thanks in advance

Just wanted to note that, according to the meter, I mixed about -52 dB
of noise back into the "noiseless" result.

Sounds more natural ?

Long ago (ca 1985) I was working on a film for Sting. I had him in the
studio to fix a couple things, one of which was tape hiss on a Space Echoed guitar track, He told me, "Leave it in, it reassures people." That has
since turned out to be quite a prescient observation. Among other things telephone companies have done a lot of work with adding noise in digital
paths, for the very same reason,

Billy Y..
--
rol -(sp) ; save carry
.purge #lun.sr ; dump open .cstat channel
ror (sp)+ ; pop carry
--- Synchronet 3.21a-Linux NewsLink 1.2

From Newsgroup: rec.audio.pro

Den 13-09-2025 kl. 06:09 skrev Harvey Sanenbum:

On 9/13/25 1:04 AM, Harvey Sanenbum wrote:

Not sure if anyone still monitors this group, but I have a question if
so.

I have been using some rather extraordinary online models and
algorithms to denoise some wildlife audio recordings I made several
years ago.-a The most startling thing I have found is how WELL these
models do.-a About half of my recordings I considered unusable.-a Low in
the noise and if software NR was applied, there were too many
artifacts.-a No longer!

So, as I'm not used to having completely noiseless results, I wonder
how much noise I should mix back in?-a Two channels are created when
using the online denoisers: the denoised track (and I mean denoised!!)
and just the noise it removed.

I come from an age of cassette tape which, even with Dolby NR, there
was always noise.-a Even when I used my software NR, I always mixed
some back in to help hide artifacts.

Just some general suggestions welcome.-a Thanks in advance.

Just wanted to note that, according to the meter, I mixed about -52 dB
of noise back into the "noiseless" result.

IF you mix noise back in, then REDO FROM START with less noise reduction applied so that you do not get any artifacts and avoid the trap of
loosing room tone as they did with Beatles "Love", always a magnificent example of what not to do.

- Peter Larsen

--- Synchronet 3.21a-Linux NewsLink 1.2