AC3 to FLAC - Shows Loss of Quality?

[TheLastOfThem]

I don't know if this is true..or if anyone has experience. I have a AC3 2 channel 192 kbps file - 16bit depth. And I saved it as a FLAC for a project...and I noticed in the spectrogram there's....all these little bits of......noise extending to 22kHz....that was NOT in the original AC3 spectrogram.

When I save to 24-bit however...these are not there. Can anyone help me as to whats going on?

[JVRaines]

What software did you use to make the FLAC file?

[TheLastOfThem]

Audacity. Went from AC3 - to FLAC. Spectrogram shows bits of noise that wasnt there before. If I do 24-bit FLAC - nothing. Just like original

[jagabo]

Set Audacity to work internally in 16 bit PCM and the same sample rate as your source (48 KHz?). And turn off dithering. That should get you cleaner spectograms -- but probably worse sound.

[TheLastOfThem]

Sample rate was 48kHZ - dithering off - got rid of the distortion to FLAC 16bit.

Why is this? I just want it to retain the full quality from the AC3? No more no less. Why would it have worse sound if the spectrogram is clean?

Unbelievably confused now how to keep authenticity of the original source - if I plan on editing it too....

I believe audacity is editing files in 32-bit depth which gets truncated to 24-bit depth on export to flac??? Not sure about anything honestly

[hello_hello]

Lossy audio doesn't have a bitdepth. If a program is telling you AC3 is 16 bit it's lying, or telling you how it thinks the audio should be decoded.

As lossy audio has to be decoded to a fixed bitdepth, the greater the bitdepth the more accurately it's decoded. There's less "rounding" to fixed values and therefore less quantisation noise. Dithering is explained here. https://en.wikipedia.org/wiki/Dither#Digital_audio

It's largely theoretical though. Decoding to 16 bit or 24 bit with or without dithering will probably sound exactly the same.

[pandy]

Sample rate was 48kHZ - dithering off - got rid of the distortion to FLAC 16bit.

Why is this? I just want it to retain the full quality from the AC3? No more no less. Why would it have worse sound if the spectrogram is clean?

Unbelievably confused now how to keep authenticity of the original source - if I plan on editing it too....

I believe audacity is editing files in 32-bit depth which gets truncated to 24-bit depth on export to flac??? Not sure about anything honestly

At first i assume this was not dithering but noiseshaping (dithering is not perceived usually differently than marginally increased noise floor), secondly - dithering is good even if spectrum looks different - in fact dithering minimize perception of distortions (decorelate distortions from signal at a cost of minimal reduction signal to noise ratio).
If you plan editing signal - use higher bitdepths than 16 bit - 16 bit is a rendering bitdepth - instead 16 bit depth use 24 bit int at least or float format.

[jagabo]

Why would it have worse sound if the spectrogram is clean?

For the same reason dithering reduces posterization artifacts in video. But as others have pointed out, unless you are under 20 and have golden ears you probably won't be able to tell.