Topic: DAC - larger than 24-bit wordlenghts sources

Hi I've been researching a bit and i couldn't find a official answer.

I wanted to know how the RME UCX performs wordlenght conversions from sources higher than it's native 24bit DAC and how does the ASIO driver works best on this subject. In other words, how should i feed it?

Does it truncates? If so, If working on a 64-bit daw, would it be better to apply a 24 bit dither before the DA for playback?

I feel this is a very sensible and always controversial subject and im really interested in knowing how the converters internally work best.

Thanks,
Ignacio

Re: DAC - larger than 24-bit wordlenghts sources

Ignacio, can you give an example of a 64-bit daw?

Do you mean daw software that handles a 64-bit word lenght PCM audio file?

What A-D converter would you use to input such a file into the daw?

3

Re: DAC - larger than 24-bit wordlenghts sources

This is a misunderstanding. If you have a 64 bit DAW then the DAW will take care of the steps needed to feed any interface with either 32 or 24 bit data. Nothing to worry about.

Regards
Matthias Carstens
RME

Re: DAC - larger than 24-bit wordlenghts sources

The PCM files are 24 bit, however due to Reaper's x64 architecture the audio path is fully 64 bit. Once the DAW catches the 24bit audio files their bandwith is extended until the end of the path.

What i want to know is how the RME UCX handles 32 bit audio data if its DAC works at 24bit. Does it truncates to 24 bit when receiving 32bit from the DAW?

Thanks!

5

Re: DAC - larger than 24-bit wordlenghts sources

Sure.

Regards
Matthias Carstens
RME

Re: DAC - larger than 24-bit wordlenghts sources

The PCM files sent to your audio interface will generally be 24 bit. The DAW will not generally send 32-bit floating point data to your audio interface. The internal mixer of the DAW program may mathematically operate using either fixed 24-bit maths, 32-bit floating point maths, or even 64-bit maths to do its internal calculations and improve its internal accuracy, however it will output a 24-bit signal to your audio interface. Some audio programs use internal dithering which happens behind the scenes, or can be switched on and off, however this is nothing to be concerned about, and is generally inaudible, as the combined effect of all digital summing errors when using 32-bit floating point maths or fixed 48-bit or 64-bit summing is completely negligible. The ASIO/CoreAudio driver receives the 24-bit output streams of digital audio data and sends them to your desired audio outputs (including using your TotalMix digital matrix to route/mix signals to particular destinations if desired). The audio-interface driver also manages the size of your audio buffer, which will affect the round-trip latency of audio signals coming into and out of your DAW. It will also manage the settings for your clocking setup, ie will your audio interface be the master or will it slave to incoming digital audio data. Do not worry about applying dithering unless you want to export a 16-bit file for final delivery via an audio CD. In this situation, apply dither once only when reducing bit-length from 24 to 16 bits. This will help minimize the affect of quantization noise, especially at low volume levels, when reducing bit-length. I generally apply dithering as my last post-fader insert on my master buss when doing a final audio export to a 16-bit file to be burnt as a WAV file for an audio CD, or as a 16-bit MP3 file to minimize file size. If you are exporting your audio data for further processing (ie exporting tracks to be passed on to a mastering engineer), leave the audio files as 24-bit files. The dithering only needs to be applied at the last stage, when the final bit-reduction for final 16-bit audio delivery takes place. These days, iTunes and other online music delivery portals will also happily accept 24-bit uploads which are then converted to various compressed formats (MP3, M4A, WMA etc...).

Kind regards,
George

Re: DAC - larger than 24-bit wordlenghts sources

ignaceland wrote:

The PCM files are 24 bit, however due to Reaper's x64 architecture the audio path is fully 64 bit.

This is a misunderstanding/misconception. The program architecture has nothing whatsoever to do with internal audio resolution. The Reaper website explicitly states "All versions of REAPER have a full-resolution 64-bit internal audio signal path from end to end, on all operating systems".

Also the program does not "extend file bandwidth", the 64 bit claim applies to internal calculations (mixing/effects/plugins), not to file handling.

Regards
Daniel Fuchs
RME

8 (edited by ignaceland 2013-08-20 11:08:47)

Re: DAC - larger than 24-bit wordlenghts sources

Thank you George and MC.

Daniel Fuchs,
My main question was not about how REAPER works and im not an expert on it at all, it was how the RME UCX works with a signal i feed. You are turning the question to other side. I preffer if you take that time to answer this question http://www.rme-audio.de/forum/viewtopic.php?id=18137

I may have expressed myself badly, but you did your best to understand what you wanted. 24 bits recorded FILES will not have more than 24 bit resolution internally, why would them?. And I know that FILES themselves are not being converted by the DAW, but while they are IN the DAW as soon as i move a fader or call a plugin or something, this file's audio will adapt (or the term you like best) to the software internal's bandwidth. This isn't something special or unknown at all and you can try it yourself freely with this little tool http://www.stillwellaudio.com/plugins/specials/bitter/ and any DAW or plugin. So why don't you answer my question instead of quoting things that i didn't mean?

I will give you a more common scenario, so you don't get lost.
I record audio with a 24-bit ADC. Then i use a software to CONVERT the files to 32-bit (now it has 8 more zeros) and i throw them into any 32bit DAW (say REAPER x86) so this is a nonquestionable 32 bit path (from a file perspective). then i apply some reverbs and those 8 zeros may be filled. What I wanted to know is what happens when this lovely path arrives at 32 bit from a DAW that keeps this 32-bit output (doesn't truncate or dither) to the UCX. How does she handles it?

That was it. You could have guided me here http://www.rme-audio.de/forum/viewtopic.php?id=10617

Im sorry but i felt you were arrogantly avoiding my question.
By the way, i really like my RME interface and I'm fully in love with her.

9

Re: DAC - larger than 24-bit wordlenghts sources

Well, you come up with a clearly wrong wording and thus statement, he gives the short and correct answer as addition to the already given information, and you think this is 'arrogant'? You wasted a lot of lines to tell us that. Sorry, but this is no internet cafe or studio lounge and as everything seems to be said - thread closed.

Regards
Matthias Carstens
RME

Re: DAC - larger than 24-bit wordlenghts sources

All your questions regarding truncation were already fully answered. My main point was to clarify that 64-bit internal audio processing is not connected to the program's x64 architecture - the 32 bit versions are no different in that respect.


Regards
Daniel Fuchs
RME