24 bit/16bit wav Render to Mp3 Quality Difference.

[vince1440]

If am taking a .WAV mixdown rendered from Ableton and immediately converting it to mp3 to send out to people, will converting to mp3 from 24 bit result in a higher quality sounding mp3 than converting from 16 bit to mp3?

I know that a 24bit wav render sounds slightly better than a 16 bit render. I tested that out extensively. I can hear the difference. It doesn't sound loads better, but 24bit does sound better than 16 bit.

The question is then, does the slight quality gain of a 24bit wav file transfer over to mp3?
When I'm converting from 24bit wav to mp3 there is no mention by my encoder that it is being dithered to 16bit first.

So you can convert 24bit straight to mp3 with no go between?
Is there a quality difference?

I can't find the answer to this.

Note: the sample rate will be 44.1 for both 24bit/16bit

[Tarekith]

If you have access to the 24bit file anyway, then I would probably use that myself just to be on the safe side. I'd be surprised though if there was an audible difference between a 24bit source and 16bit source by the time it gets to MP3.

Since you're one of the very few people I've heard of who can hear the difference between 16 and 24bit wav files, I'd say trying it both ways and comparing it on your own is worth a try too.

[vince1440]

Only one reply anyone else wanna give their 2 cents?

[Stromkraft]

Only one reply anyone else wanna give their 2 cents?

Tarekith isn't enough?

I'm not sure what to say, but I'd of course test by nulling the 2 MP3s and listen to the differences, if any.

[102455]

From Google - http://taperssection.com/index.php?topic=136471.0

[Angstrom]

From Google - http://taperssection.com/index.php?topic=136471.0

From That thread this comment sums up my thoughts, although its a bit hyperbolic

Converting a 24-bit lossless file directly to mp3 can be downright dangerous, because you're relying on the mp3 compressor to convert to 16-bits (there is no such thing as a 24-bit mp3).

Depending on what program you use it may not dither, but rather truncate (chop off) the top 8 bits, thus doing very bad things to the sound. Obviously good editing programs can do the job right if given the right instructions, but your average "converting tool" can screw things up big time. Some of these programs are notoriously bad at sample rate conversions too.

[Stromkraft]

Only one reply anyone else wanna give their 2 cents?

Tarekith isn't enough? 8O

I'm not sure what to say, but I'd of course test by nulling the 2 MP3s and listen to the differences, if any.

OK, I made some tests. First I failed to null* an exported 32bar section from my own song, when I polarity inverted the 16bit version, exported as non-dithered and non-normalized from Live, playing back at the same time as the 24bit version inside Live. This was somewhat not surprising, but it's not in line with the results I got making a 16bit version from the 24bit version with XLD, which did null 100%*. Likely exports are never exactly the same depending on source. After all it was done with MIDI instruments in my case. I have to repeat with audio only (Gaaaaah!).

Export settings in Live (only difference being 16bit vs 24bit):



I decided to use the 24bit version as the master and make a 16bit source from that with XLD:



This is more how I'd done otherwise too. I made these mp3 conversions:

• In XLD I used these settings for the LAME MP3 encoder
  
  and made one mp3 file directly from the 24bit master.
• and another from the 16bit version

Note that XLD, in both cases made a 16bit version, whatever that may mean with MP3 files. I also tried with iTunes 12.6.1.25 and Sbooth Max 0.9.1 and as far as I can could deduct the result is viewed as a 16bit MP3 file. There may be other tools that can do more floating bits, but I'm unaware of them.

The Results
When I attempted to null* these MP3s in Live — which means both files, the 24bit source and the 16bit source, got decoded to lossless format — I managed to get almost 100, perhaps 99.something% with only some MP3 fizzle being heard. While clearly being somewhat different, I seriously doubt anyone can hear a difference between these two paths of MP3 encoding.

Nulling details
*Nulling was done inside Live as my normal tool Ocenaudio didn't behave. Both tracks were at 0dB and Utility with preset "Phase invert" was used. Clips were unwarped.

I didn't adjust starting point, but worth noting that most MP3 encoders inserts, intentionally or not, what I believe to be a non-standard length of silence at the beginning of the file. As I said above the LAME MP3 encoding is used in XLD. This silent beginning makes it difficult to reliably attempt to null an MP3 with the AIFF/Wav original without adjusting the starting point of the clip. This operation can be done also in unwarped files and is necessary in order to make a meaningful comparison between the lossless original and the lossy decoded MP3 encoding with a nullification attempt. At best position you will hear what is missing from the MP3 encoded file, or at least what is different, between the two.

[vince1440]

Thank you Stromkraft for your experimentation. I appreciate it. I also just did a test and while I can clearly here a difference between a 16/24 bit wav file, I cannot hear any difference when encoded to mp3. I will just convert 16 bit to mp3.