exporting 8bit audio that doesnt have "noise" added

Audacity 3 Help Forum Windows
1

I realise that 8-bit audio can still sound almost identical to 16-bit audio since theres over 65,000 different volume levels of audio each bit can be at between -inf db and 0db
but whenever I try to export in 8-bit I can only use the large wav audio format and NIST-SPHERE for signed 8bit PCM and it adds more-than-noticable soft-noise in the background
MP3 doesnt offer 8bit export option

When I import it back into audacity to remove the noise.
The only option is to do the noise removal/reduction and set frequency bands to max and use default options (I used the beginning of the song and copy-pasted a few 10ths of a second of noise in order to be able to get noise profile
then i removed the noise and removed the copy-pasted second of empty-samples

2

Is there a question in there somewhere?

If you’re asking, “is noise unavoidable with 8-bit audio”, the answer is “yes, unless the audio is total silence”.
If you’re asking how to make extremely small audio files while preserving as much audio quality as possible, then look at Opus (https://www.opus-codec.org/), iLBC (https://webrtc.org/license/ilbc-freeware/), AMB-NB (http://wiki.multimedia.cx/index.php?title=AMR-NB) and G.711 (https://en.wikipedia.org/wiki/G.711).

3

Forgot to note my operating system is windows 7 ultimate 64-bit incase that has some codecs/formats changes (couldn’t edit post)
And I wish to use 8-bit format for some audio files in a reasonable format with minimal compression but not so large file size and if possible export in a format that doesnt add too much “noise” to the sound from 8-bit format limits

4

I still don’t understand the question, if there is a question. If you want to avoid the noise that is inherent in 8-bit format, then don’t use 8-bit format. FLAC format will give approximately the same file size as 8-bit WAV but with vastly superior sound quality (the same sound quality as 16-bit WAV). If you want smaller file size than that, then MP3 or OGG will give better sound quality than 8-bit WAV.

5

I realise that 8-bit audio can still sound almost identical to 16-bit audio since theres over 65,000 different volume levels of audio each bit can be at between -inf db and 0db

With an 8-bit variable, you can only hold values between zero and 255. So no… You cannot get quality that’s almost identical to 16-bits.

but whenever I try to export in 8-bit I can only use the large wav audio format and NIST-SPHERE for signed 8bit PCM and it adds more-than-noticable soft-noise in the background

[u]Quantization Noise[/u]. It’s most noticeable at low signal levels (and at low-resolution), but unlike analog-noise it’s non-existent with dead-digital-silence.

MP3 doesnt offer 8bit export option

Right… MP3 uses floating-point. File size is determined by the bitrate (kbps = kilobits per second). There are 8 bits in a byte, so you can divide by 8 to find file size in kilobytes per second.

And I wish to use 8-bit format for some audio files in a reasonable format with minimal compression but not so large file size and if possible export in a format that doesnt add too much “noise” to the sound from 8-bit format limits

Maybe [u]ADPCM[/u]? ADPCM is common in telephone systems.

Lowering the sample rate (kHz) will reduce the file size (of an uncompressed file) proportionately. i.e. A 22,050Hz file is half the size of a 44,100Hz file. The “cost” is loss of high-frequency information. The audio can only contain frequencies up to half the sample rate… An 8kHz “telephone quality” file, can only contain audio up to 4kHz.

with minimal compression

When you make the file smaller, that’s compression. Unless you are using lossless compression (FLAC or ALAC)* you are throwing-away data when you make the file smaller and there is potential quality-loss. When the quality loss becomes noticeable, it becomes a compromise.

If you are compressing speech, there are compression algorithms optimized for speech.



\

  • With lossless compression, you can typically get a file 60% of the original-uncompressed size. Apparently, you need more compression than that.
6

Actually, 8 bit only has 256 levels… :stuck_out_tongue:

Astounding it can still sound OK on some audio, isn’t it?

The biggest disadvantage with 8-bit is the noise floor, at about -48 dB. You need about 60 dB for music that has normal dynamics, so you’ll have to compress the hell out of it. And if you think you can still make a good-sounding mp3 out of an 8-bit audio file, you’re in for a world of disappointment… :mrgreen:

If quality isn’t the main concern, you could look at the speex codec:
http://speex.org/
It’s in ffmpeg, and while it is for speech, I have used it for low quality music, like for VOIP answering machines. It’s fast and compresses well.

7

Just to clarify… cyrano is talking about dynamic compression, which is totally different from file compression (MP3, etc.).

8

Just to clarify that one, ADPCM is 4-bit.


Gale

9

When I said 8bit audio vs 16-bit audio I was meaning the 16-bit audio had over 60,000 levels not the 8-bit audio

8-bit is 256 levels of course