Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[eut]

While playing back 192kHz/32 bit files I realised that the playback is inaccurate in such way that for instance one of two guitars is relegated almost entirely to a faint background sound whereas in the original recording it is quite prominently reproduced on one stereo channel. It took me a while to find out that MC15 does not accurately reproduce my 32bit/192kHz files, as it moves certain sounds/instruments of an entire song into the background. Initially I suspected my sample rate conversion software, but after playing the 32bit/192kHz wav files in Foobar they were reproduced in a perfectly normal way.
Can there be a problem with the MC15 settings?
Any assistance is much appreciated as I really like the program. I only recently emigrated from years of Foobar use to MC 15, but it seems that I am now back to Foobar.
Ernst 

[JimH]

Are you playing to a receiver?  If so, try changing the playback mode on it.  2ch Stereo, for example.

[eut]

I am playing to a masterclock via firewire output and I am in 2 channel stereo output mode

[Alex B]

The playback engine in MC appears to handle 32-bit signals correctly. I just did a little test. I played three 192 kHz files using MC's Diskwriter mode - 24-bit integer, 32-bit float and 32-bit integer. I created the test files with Adobe Audition (24-bit and 32-bit float) and Sony Sound Forge (32-bit integer). MC's output bit depth was set to 32-bit (this creates 32-bit float wave files). I compared the resulting files with the original source files by creating difference files in Audition (copy the source file-> invert-mix-paste on top of the destination file). The played 24-bit integer and 32-bit float files were identical with the original files (the difference files were completely silent). There was a faint difference between the resulting 32-bit float file and the original 32-bit integer file, which was expected because the storage format changed. The amplitude of the difference file was about 200 dB below 0 dBFS:

Code: [Select]

Left Right
Min Sample Value: 0 0
Max Sample Value: 0 0
Peak Amplitude: -186.6 dB -186.62 dB
Possibly Clipped: 0 0
DC Offset: 0 0
Minimum RMS Power: -220.29 dB -222.29 dB
Maximum RMS Power: -188.35 dB -188.25 dB
Average RMS Power: -204.6 dB -204.85 dB
Total RMS Power: -202.66 dB -202.51 dB
Actual Bit Depth: 1 Bits 1 Bits

Using RMS Window of 20 ms
So even in this case the difference is totally insignificant.

It would also be very odd that something would alter the signal only partially like you explained. Somehow something after MC's playback engine is causing the difference you are hearing.


It might help if you could list your exact output mode and DSP settings, volume settings (are you using the internal volume option in MC?), your OS and your sound device details. Also, you have not explained why you are using 32-bit files. How did you create them and what is the exact storage format? The output devices don't usually handle more than 24 bits (some drivers need 32-bit packages, but that is different, the DAC still won't process more than 24 bits.)

[eut]

Many thanks for the indepth analysis and the tests.

It would also be very odd that something would alter the signal only partially like you explained. Somehow something after MC's playback engine is causing the difference you are hearing.

I agree that it would be very odd if something is only partially altering the signal, though I have experienced a similar problem also with another prograrm.
Just to add more to the confusion, 32-bit 96kHz wav files I have upsampled to 32bit-192kHz using Cirlinka's HD-Audio Remaster 3.2 software (I usually upsample with Voxengo's r8brain) do not play at all on MC15 and produce only a high frequency hissing sound. The same files play perfectly well on Foobar, which indicates that there is no hardware issue at play.

It might help if you could list your exact output mode and DSP settings, volume settings (are you using the internal volume option in MC?), your OS and your sound device details. Also, you have not explained why you are using 32-bit files. How did you create them and what is the exact storage format? The output devices don't usually handle more than 24 bits (some drivers need 32-bit packages, but that is different, the DAC still won't process more than 24 bits.)

I will post details of my MC settings as soon as I am back at my home computer. 
Now I only remember that system volume settings are at 100% and that I am not using internal MC volume settings. My PC is a Sony VGN-Z58GG running on Windows 7 Pro 64-bit with an internal soundcard. I do route my sound directly through a firewire 400 connection into a Brainstorm Electronic Distripalyzer masterclock and from there into an EE Minimax DAC with Sabre ES9018 32-bit chip that accepts 32-bit signals.
I am using 32-bit files (wav) as I have experienced that they sound better in my system than 24-bit files. So far I have only played back 32-bit 96kHz wav with MC (without any problems) as my previous DAC (Cambridge DAC Magic) did not accept higher than 96kHz signals. The new DAC accepts 192kHz input, hence my upsampling to 32-bit 192kHz which appears to cause the problem with MC.
More details on my MC settings will follow.

[eut]

My MC settings are as follows: Output mode is ASIO with 4.00 sec buffering and volume device set to my Brainstorm DCD-8 Audio. No DSP settings are enabled.

[Alex B]

So you are playing the files without DSP directly to the ASIO driver.

I think it would be useful to try a sample file. I uploaded a short 96 kHz 32-bit float sample file here: http://rapidshare.com/files/436765641/96_32float_sample_wav.zip (about 10 MB)

Could you please download & unzip the package and try playback? First the wave file as it is. Then resample and save the file using your two different methods. Try also the 24-bit and 32-bit output modes in MC's settings (DSP Studio > Output > Bitdepth).

If you can reproduce the problems with the resampled versions, upload the files in a zip package to some file sharing service and post the download link here (www.rapidshare.com, www.megaupload.com, www.sendspace.com, etc). Then we can try if the problem can be reproduced on other PCs and also the JRiver developers can test the samples.

Naturally, if the odd channel coupling/phase problem occurs only with specific content (but not with my sample) we would need to have a sample of that content.

[eut]

Many thanks for the file Alex, which was a bit too short to reproduce more than subtle changes in an 196kHz version upsampled by Voxengo's r8brain. I still seemed to have heard a very slight difference. Upsampling with Cirlinka's HD-Audion Remaster however produced an inaudible file, which I will be sending you shortly. I will also send copies of one 'original' 96_32 wav file and the 196_32 version upsampled with r8brain, where the difference is strikingly clear, as nearly the entire brass section is missing. It will just take a while to upload these. Once I have the link I will post it.
Ernst

[eut]

Here is the link for first two files
http://www.filemail.com/dl.aspx?id=MZBITPSDWZOCAOT

[eut]

And this is the link to the reference 96_32 wav file
http://www.filemail.com/dl.aspx?id=WEGMBBHHSQWWGUU

[Alex B]

Upsampling with Cirlinka's HD-Audion Remaster however produced an inaudible file, which I will be sending you shortly.

You are correct. The wave format is incompatible with MC, even though some other programs can decode it correctly.

To Matt (from JRiver): it is the latter, ~20 MB, sample file in this link: http://www.filemail.com/dl.aspx?id=MZBITPSDWZOCAOT

I will also send copies of one 'original' 96_32 wav file and the 196_32 version upsampled with r8brain, where the difference is strikingly clear, as nearly the entire brass section is missing. It will just take a while to upload these. Once I have the link I will post it.

Both files appear to play correctly on my computers. I couldn't hear any audible differences between MC's and foobar's output. In addition I repeated the "play to Disk Writer" test with the files. The resulting output file of the 192 r8brain version was identical with the source file. The difference file in Audition was completely silent.

I noticed a small difference in the 96 kHz version when I compared the Disk Writer output with the source file. Its phase was reversed. However, after fixing the phase reversal, also these two files were identical. The phase was reversed in both channels so it should not cause any practical problems.  (Matt, it appears that 32-bit integer files cause this behavior. The sample is this: http://www.filemail.com/dl.aspx?id=WEGMBBHHSQWWGUU ).

Since you are hearing a clearly audible difference when you play the 192 kHz 32-bit file even though MC can decode it correctly to a disk file, the problem is obviously caused by some kind of incompatibility with the device driver or hardware.

Can you reproduce the problem if you force MC's output to 192 kHz and 32-bit in the DSP options? Does the problem disappear if you set the output to 24-bit? You can use a lower resolution file for testing this.

[eut]

Both files appear to play correctly on my computers. I couldn't hear any audible differences between MC's and foobar's output. In addition I repeated the "play to Disk Writer" test with the files. The resulting output file of the 192 r8brain version was identical with the source file. The difference file in Audition was completely silent.

I noticed a small difference in the 96 kHz version when I compared the Disk Writer output with the source file. Its phase was reversed. However, after fixing the phase reversal, also these two files were identical. The phase was reversed in both channels so it should not cause any practical problems.  (Matt, it appears that 32-bit integer files cause this behavior. The sample is this: http://www.filemail.com/dl.aspx?id=WEGMBBHHSQWWGUU ).

Since you are hearing a clearly audible difference when you play the 192 kHz 32-bit file even though MC can decode it correctly to a disk file, the problem is obviously caused by some kind of incompatibility with the device driver or hardware.

I was afraid of this, but realised very recently that the small indicator bars on the very left and right side of the uppermost window in MC, which are active when a tune is played, did indicate activity in the right channel at the beginning of one of the tunes that was missing a guitar, while there was nothing coming out of the speaker that should have played the guitar sound.

Can you reproduce the problem if you force MC's output to 192 kHz and 32-bit in the DSP options? Does the problem disappear if you set the output to 24-bit? You can use a lower resolution file for testing this.

I did try the 192kHz and 32-bit DSP option and also the 192kHz 24-bit option, both with the same result, some information are missing. When I play 96kHz 32-bit, I have no problem with a perfectly correct rendering.