Hey there,

I've found some old speech-recordings which where encoded as non-joint-stereo 128kbit CBR mp3s, although the source was just mono. Is there a way to reduce these files to 64kbit mono mp3s without reencoding them?

Greetings,
Dr.Retro

same question here!

not possible but even if you do re-encode, given that it's just speech, the quality doesn't matter that much and wouldn't deteriorate that much further.

uh... man. i would have gotten 10 warnings already for saying what you did.

as i understand it, for many codecs... and most people's perceptions of artifacts... speech is actually one of the hardest things to encode properly.

I believe that would be possible, but you might also want to consider reencoding with a codec designed for speech. You could end up with much less than 64 kbps.

I have a 192kbps CBR simple stereo mp3 that only has audio in one channel. I would like to split the mp3 into two mono files without re-encoding. It's mostly for aesthetic reasons rather than saving space, as I'd rather here the mono signal in both speaker/headphones rather than one.

Now, searching around I found two other threads that suggest that this is theoretically possible, but nobody's aware of a repacking tool that can do it.

http://www.hydrogenaudio.org/forums/index....showtopic=40808
http://forum.videohelp.com/topic307988.html

Is that still the case? How hard would it be to make such a tool?

I reckon it's so rare for such a thing to be useful (i.e. you don't have a lossless source to rework and re-encode) and the gain from not transcoding is modest, that it's not worth the fair amount of effort involved to create a general tool for this.

If you could get at the sources of that mp3 repacker or something similar (is it open source?), you might get somewhere with relatively little extra work if it matters enough to you, and you'd then get a 96 kbps Mono MP3 I presume.

The alternative is to decode to WAV and remove the silent side or duplicate the channel that contains audio, then transcode in whichever way seems best.

For maximum compatibility you could re-encode using something like LAME VBR (perhaps -V1 to reduce transcoding artifacts caused by unmasking, though temporal smearing could be noticeable). If it's mono, it'll probably come out much less that 192 kbps.

If it's suitable, avoid transcoding by saving to lossless, such as FLAC, Wavpack or Apple Lossless (if you use an iPod without Rockbox). Mono 16-bit files should come in at 450-750 kbps, I'd imagine even if presented as 16-bit/44.1kHz stereo. Perhaps you'd only then transcode when your target device supports only lossy.

Unless your material has a low-pass filter below 16 kHz, I'd also suggest LossyWAV pre-processing (e.g. -q 5) as an option to reduce your FLAC or WV bitrate (but not Apple Lossless) further with very few transcoding artifacts and no increase in temporal smearing. You might (I guess) get to 300-400 kbps or even lower at -q 5 and perhaps even around 200-250 kbps at -q 0, so you'd barely lose out on space used and have a file much more pleasant to listen to on headphones.