Normalizing mp3 and ogg tracks to 95 dB

[rowjekto]

Is there an accurate way for it?

[hello_hello]

If you mean ReplayGain normalising, MP3Gain will do it for MP3 or AAC as it lets you adjust the target volume (normally 89dB). 95dB would be too loud for some audio as it'd increase the volume enough for the peaks to be clipped a little on playback. If you don't mean ReplayGain normalising, I don't know what 95dB means.

There's no way to adjust the volume for ogg losslessly that I know of. You'd need to scan with ReplayGain, save the data, then convert while applying the appropriate volume adjustment.

Foobar2000 can also adjust MP3 and AAC losslessly as MP3Gain does, or it'll use existing ReplayGain tags to adjust the volume when converting. You could enable ReplayGain +6dB in the converter configuration and it'd convert to a target volume of 95dB. You'd probably want to scan the output file and check for clipping though.

[rowjekto]

MP3Gain will do it for MP3 or AAC as it lets you adjust the target volume (normally 89dB).

MP3Gain isn't accurate: it's limited to 1.5dB steps mode.

If you don't mean ReplayGain normalising, I don't know what 95dB means.

It means volume level that is equivalent to ReplayGain's 95dB.

There's no way to adjust the volume for ogg losslessly that I know of.

It doesn't have to be losslessly.

Foobar2000 can also adjust MP3 and AAC losslessly as MP3Gain does, or it'll use existing ReplayGain tags to adjust the volume when converting.

Is foobar2000 accurate? Isn't foobar2000 gain limited to its player?

[hello_hello]

The 1.5dB step limitation is an MP3 limitation.

If it doesn't have to be lossless, then yes foobar2000 will do it accurately. The converter uses command line encoders and will decode/output to 32 bit float if an encoder supports it. The player isn't involved in the conversion process, although they both decode with ffmpeg.

I have the job of putting the music on my mother's MP3 player. She uses it to teach line dancing in a hall with a small sound system so I have to squeeze as much volume out of it as I can. I chose 93dB as the number of tracks that'd clip was still reasonably low. I adjust the volume losslessly if it's possible without clipping, but for conversion I created a preset with a limiter DSP in the conversion chain.

If you convert to a lossless format and scan the output files you should end up with exactly the volume you specified, aside from any deviations that might be caused by clipping or limiting. Lossy codecs tend to deviate by a small amount, but that's what lossy codecs do.

Converting to 95dB. FLAC first, AAC second in the screenshot below. Oh.... and Foobar2000 uses the EBU R128 scanning algorithm rather than the original ReplayGain algorithm, which is probably a bit more accurate, so the result can be a little different if you scan with MP3Gain. Mostly they tend to agree to within about 1 or 2 dB, if memory serves me correctly.

The track I used for converting was originally scanned with MP3Gain so I scanned it again before converting it. MP3Gain had saved a TrackGain of +0.06dB to the tag. Foobar2000 said the track gain is -1.55dB, so in that case the difference between scanning algorithms is around 1.5dB.... I thought I'd mention it in case that bothers you.

[rowjekto]

Thanks for the detailed answer.