Cics - Cplay and Cmp2

[SunRa]

Well, I should say that I am now using cPlay through some lousy desktop speakers and I am definitely hearing something else than using foobar. In fact I've tried foobar with virtually any output (asio, driver, wasapi) and various sample rates, but I could not distinguish some significant changes.. Maybe when changing to asio and buffering in RAM I had some subtle changes.

But it's nothing comparing to what I am hearing with cplay...

Just for reference I am using win7, not xp, 32 bits.

[phofman]

SunRa,

did you check if both foobar and cplay are bit perfect? The article http://www.enjoythemusic.com/magazin...chapter106.htm says:
RAM loading of a 640 mB CD takes about 15 seconds from the hard drive, and the information is then upsampled by the program to the highest allowable amount for your DAC or 192 kHz.

I cannot comment on that, I do not know the internals.

Playback on windows is a black box to me due to missing source code of the whole chain from the player (I could not find complete source code for cplay or foobar2000) through the library layer down to the drivers. Any layer can modify the audio stream which can be detected only by checking for bit-perfection.

[SunRa]

Hello,

thanks for the reply. Do you have an easy way to check for bit perfect? I really haven't any time for looking on this but if you have an easy ready made method I can check this week-end.
I plan to compare a file up-sampled at 192/24 played through both players. Also, I plan to compare the 192/24 played through foobar with the original 44.1/16 played through cPlay and realtime upsampled with src or sox (cplay can upsample using different src and different sox configurations).

Thanks!

[phofman]

Well, there is the simple test of playing a DTS file through the chain and SPDIF output, checking if an AV receiver can detect the incoming stream as DTS. I am not very convinced about conclusiveness of the test since it is rather simple to check for DTS header and the SW can reconfigure the chain accordingly.

I guess the only bulletproof test is recording the SPDIF output with SPDIF input of the same/another card and compare the results. Time-aligning of two wavs in audacity takes a few minutes, the rest is just subtracting the samples of the two wavs and checking the result for non-zero samples, e.g. in sox.

This method is certainly not a simple one. That is the reason not many people actually check for bit-perfection.

[phofman]

Takashi Iwai, administrator of alsa, talks about latency

http://mailman.alsa-project.org/pipe...er/021107.html

[SunRa]

Hello phofman,

thanks for the tips, I'll see what my card can do... the part with subtracting from wav it's not very clear to me, I'll play with audacity.

Now, does anyone know how can I switch off upsampling in cPlay? Let's say I have 192/24 files, if I set 192Khz upsampling with sox in cPlay, does it mean than when detecting a 192 file it won't intervene on it?

[phofman]

Originally Posted by SunRa

Hello phofman,
thanks for the tips, I'll see what my card can do... the part with subtracting from wav it's not very clear to me, I'll play with audacity.

The goal is to compare the outgoing and incoming wavs. They have to be time-aligned first (e.g. in audacity) to start with the same sample. Then you have to compare them somehow. For me the easiest way is to subtract the two wavs in sox and check the resultant wav statistics:

sox -V -m outgoing.wav -v -1 incoming.wav -n stat

Params:
-V - verbose output
-m - merge input wavs
-v -1 - volume -1 for the second input wav, i.e. multiply its samples by -1, i.e. invert them. Merging thus means subtraction
-n - do not create output wav
stat - calculate and print statistics for the resultant wav (product of the subtraction). If the two wavs are identical, max/min/avg samples must be exactly zero.

Originally Posted by SunRa

Now, does anyone know how can I switch off upsampling in cPlay? Let's say I have 192/24 files, if I set 192Khz upsampling with sox in cPlay, does it mean than when detecting a 192 file it won't intervene on it?

Sox itself detects when the incoming rate is equal to the requested output rate and skips the conversion.

[soundcheck]

Originally Posted by phofman

SunRa,

did you check if both foobar and cplay are bit perfect? The article http://www.enjoythemusic.com/magazin...chapter106.htm says:
RAM loading of a 640 mB CD takes about 15 seconds from the hard drive, and the information is then upsampled by the program to the highest allowable amount for your DAC or 192 kHz.

I cannot comment on that, I do not know the internals.

Playback on windows is a black box to me due to missing source code of the whole chain from the player (I could not find complete source code for cplay or foobar2000) through the library layer down to the drivers. Any layer can modify the audio stream which can be detected only by checking for bit-perfection.

He is not loading a full CD into RAM with CPLAY. But I am doing it. This way I can stop the HDD for about 45 minutes.


There is not a fixed upsampling. You just select it. You can choose Secret Rabbit or SOX.

[SunRa]

Hello,

do you guys know any application running under Windows capable of doing off-line up-sampling with sox or src?

[phofman]

Originally Posted by SunRa

Hello,

do you guys know any application running under Windows capable of doing off-line up-sampling with sox or src?

A simple DOS .bat file calling sox for each file?