Determining recording method from file (or logs)

[Cwluc]

It's hard to believe that after over a year of recording in Audacity I'm still working out some of the particulars. I have no technical issue, everything is going well with ASIO and the latest nightly build. I noticed that I forgot to turn on the external soundcard one day and it defaulted to the worse case option (my fault, not Audacity's) the internal Realtek. Once I turned the E-MU on and restarted Audacity I assumed everything was cool but it may have actually been recording with MME instead of ASIO. Is there any way to tell which method was used besides me just recording everything again?

[steve]

I assumed everything was cool but it may have actually been recording with MME instead of ASIO. Is there any way to tell which method was used

1) If the recording sounds OK, does it matter whether MME or ASIO was used?
2) Unless you built Audacity from the source code you should not have ASIO support in Audacity. ASIO support requires building Audacity with the ASIO SDK from Steinberg, but the ASIO SDK is not open source and cannot be legally distributed with Audacity (see here: http://wiki.audacityteam.org/wiki/Featu ... _Implement )

On Windows, Audacity supports MME and DirectSound.

[Cwluc]

It matters because I'm making 24 bit recordings of my records in Windows which as I understand in layman's terms MME makes a guestimation of it as opposed to the more accurate ASIO and yes I built it myself.

[steve]

and yes I built it myself.

Congratulations, that's one up on me
So far I've failed miserably building on Windows (though I generally use Linux and build Audacity regularly on that).

Yes you're correct. For 24 bit recording on Windows you (currently) need to use ASIO.

If you still have the Audacity Project, or if you have exported an uncompressed audio file in 32-bit float format, and you have not processed the recording in any way (not even amplifying or Normalizing) then it may be possible to analyse the recording to see if it was recorded in 24 bit. As far as I know there is no logged information that will tell you that, and it is unlikely to be possible to measure any discernible difference in sound quality unless the recording level was extremely low. If you do have the original 32-bit data, then you look at it with a Hex editor to see if there are sample with values that lie between the 16 bit integer values.

Personally I would only resort to recording everything again if there was a problem with the sound quality.

[PGA]

On Windows, Audacity supports MME and DirectSound.

Excuse me butting in on this topic, but what's the difference between these two and is one preferred over the other?

[Cwluc]

If you still have the Audacity Project, or if you have exported an uncompressed audio file in 32-bit float format, and you have not processed the recording in any way (not even amplifying or Normalizing) then it may be possible to analyse the recording to see if it was recorded in 24 bit.

What if I was recording in 24bit and not 32 float?

On Windows, Audacity supports MME and DirectSound.

Excuse me butting in on this topic, but what's the difference between these two and is one preferred over the other?

Did you literally mean only MME and DirectSound? Or the difference between MME and ASIO?

[steve]

What if I was recording in 24bit and not 32 float?

If you still have the original data in an Audacity Project, then I think that may still be able to analyse the data as described previously.
If you Exported as 24-bit, then the way that Audacity currently works, ALL exported audio passes through the "rendering engine" which operates at 32-bit float, so the file will be down-sampled from 32-bit float to 24 bit which by default will dither the output, creating values "in between" the 16 bit values.

I believe that it is likely that this behaviour will be changed in the future so that if there is only one audio track, then the export will be able to bypass the rendering engine, thus making bit perfect copies of integer format audio easier to achieve.

what's the difference between these two and is one preferred over the other?

MME and DirectSound are audio API's.
On Windows XP, Direct Sound can offer significantly lower latency than MME, though MME is generally a "safer" option.
On Vista and 7 DirectSound no longer offers low latency, but it does allow "exclusive mode" (if enabled in the sound system settings), so if the recording sample rate is supported by the sound card hardware then there should be no resampling by Windows between the sound card and Audacity.

I'm not a Windows expert, but I get the impression that if DirectSound works well on your system, then it is probably best to use that, but if you have problems with it, then MME is the safer bet.

[PGA]

I'm not a Windows expert, but I get the impression that if DirectSound works well on your system, then it is probably best to use that, but if you have problems with it, then MME is the safer bet.

Steve,
I note your declaration about not being a Windows expert, but I'll still trust you! I like "safer", so I'll stay with the "factory-shipped" setting of MME for my Audacity installation. Thanks for the advice.

[waxcylinder]

I am a Windows XP user and I use Windows Direct Sound on both my PCs (one is an old, low memory, desktop with XP-HE, t'other is a newish lapup with XP-PRO) and I used Windows Direct Sound quite happily on both those machines. I chose WDS over MME as it is a more moderm implementation of an API.

My wife runs W7 on her new laptop and on that she uses WDS on that.

I'd love to use an ASIO interface as my external soundcard (Edirol-UA-1EX) supports 24-bit for which ASIO is required - but I don't have the inclination to self-build Audacity

WC