ReplayGain: My approach

Post your questions & get help from friendly LibriVoxers
Post Reply
KevinS
Posts: 6952
Joined: April 7th, 2019, 8:32 am
Contact:

Post by KevinS » May 15th, 2020, 9:10 am

Not really knowing what I'm doing half the time, I present my use of ReplayGain in my recordings.

As an example, I'll speak about a recording of about 45 minutes that I am now editing. I'd say I was interrupted 4 (maybe 5 times) during the recording, so there are that many segments divided by a convenient bar/rest/whatever you call it.

I use ReplayGain on each part separately and then select the whole recording and run it through ReplayGain. There isn't often a great difference in each part and that final run of ReplayGain gives minimal adjustment.

Is there a downside to this? It doesn't sound to me that this much 'processing' has any discernible effect on the sound/quality of the recording.

MichaelMaggs
Posts: 239
Joined: April 27th, 2012, 1:10 am
Location: UK

Post by MichaelMaggs » May 18th, 2020, 7:00 am

Assuming that the recording setup (volume etc) is the same in each segment, running ReplyGain twice isn't likely to do much. You might just as well simply run it once over the whole thing.

The only issue I find is that when ReplayGain is run on a long track (say over 30 mins) it has a tendency to make the recording a little quiet. If that's the case and the result fails the Checker volume test, I then just use Amplify to bring it up to the level I want.

KevinS
Posts: 6952
Joined: April 7th, 2019, 8:32 am
Contact:

Post by KevinS » May 18th, 2020, 7:19 am

MichaelMaggs wrote:
May 18th, 2020, 7:00 am
Assuming that the recording setup (volume etc) is the same in each segment, running ReplyGain twice isn't likely to do much. You might just as well simply run it once over the whole thing.

The only issue I find is that when ReplayGain is run on a long track (say over 30 mins) it has a tendency to make the recording a little quiet. If that's the case and the result fails the Checker volume test, I then just use Amplify to bring it up to the level I want.
Thanks much for commenting! Yes, the difference in volume for each section is not far off one from the other. I suppose I've been unnecessarily fastidious (or ridiculous.)

lurcherlover
Posts: 945
Joined: November 10th, 2016, 3:54 am
Location: LONDON UK

Post by lurcherlover » May 18th, 2020, 7:29 am

To get the best sound it is good to keep the maximum Peak reading to -10dB Full Scale or even lower at -12dB. This means the recording is clean as even peaks that reach -4dB can have some small distortions. Also best to make the original recording at 24bit wav or 24 bit flac files. If really low on disk space 16 bit will do, but not so good as 24bit.

This will be below the volume level that LV wants so you will need to amplify up to that level. But as the files were recorded cleanly pushing up the gain in amplify or whatever will not degrade the sound, as the original will be clean.

KevinS
Posts: 6952
Joined: April 7th, 2019, 8:32 am
Contact:

Post by KevinS » May 18th, 2020, 9:08 am

lurcherlover wrote:
May 18th, 2020, 7:29 am
To get the best sound it is good to keep the maximum Peak reading to -10dB Full Scale or even lower at -12dB. This means the recording is clean as even peaks that reach -4dB can have some small distortions. Also best to make the original recording at 24bit wav or 24 bit flac files. If really low on disk space 16 bit will do, but not so good as 24bit.

This will be below the volume level that LV wants so you will need to amplify up to that level. But as the files were recorded cleanly pushing up the gain in amplify or whatever will not degrade the sound, as the original will be clean.
Hahah! Will Google translate work for technical language?

In reality, I think I can come to understand this. Thanks much for commenting!

lurcherlover
Posts: 945
Joined: November 10th, 2016, 3:54 am
Location: LONDON UK

Post by lurcherlover » May 18th, 2020, 12:24 pm

Sorry if it was a bit too over technical! 0dB (zero) is the point where it will clip/distort (Going into the "red")

Everything below that is minus dB (-1dB - 2dB etc). Audible.com only allow -3dB so if the highest peak reading is -3dB or -4dB it's OK - but -2dB or -1dB would be a no, no!

They also ask for a noise floor (i.e. the room sound or "silence" to be -60dB). In other words 60dB lower than the peak at zero dB or 57dB lower than their accepted peak of -3dB.

There is no such thing as absolute silence in recordings as no matter how quiet the room or studio is, there is still electronic noise (mic, recorder, pre-amp etc) as well as some "atmos." But we used to tolerate lots of noise in recordings pre digital such as tape hiss, turntable rumble, stylus noise and all those damaged LP's with scratches. But a lot of recordings made outside of the studio have "room sound" as well as birds singing in the background. I turned up my recorder gain to full volume and I could hear the birds singing and other noises such as tummy rumble and to top that my wife flushed the loo. Even at normal gain levels I can hear the washing machine when its on, way downstairs in the kitchen. It's only really quiet here when she is at work, and now work is at home, so I can't record when she's on line to students. In fact one of my readings had some piano music, probably Beethoven, as she was teaching online. But one would not wish to shut up the songbirds or Beethoven, they take precedent over recording!

TriciaG
LibriVox Admin Team
Posts: 45683
Joined: June 15th, 2008, 10:30 pm
Location: Toronto, ON (but Minnesotan to age 32)

Post by TriciaG » May 18th, 2020, 12:31 pm

Fortunately, we're not Audible. ;)
One Mystery Story Left: Master of Mysteries
Mussolini's speeches thru 1923: LINK
Samuel Smiles Self-Help: Character

KevinS
Posts: 6952
Joined: April 7th, 2019, 8:32 am
Contact:

Post by KevinS » May 18th, 2020, 12:34 pm

lurcherlover wrote:
May 18th, 2020, 12:24 pm
But one would not wish to shut up ... Beethoven, they take precedent over recording!
Especially when played by one's bride!

Post Reply