Bit-identical playback CAN sound different

[yamamoto2002]

For control experiment, it is nice to have two recorded data of the same file

[yamamoto2002]

34 minutes ago, manisandher said:

 

That doesn't sound too promising to me. The MOTU M2 is connected to the regular work PC on my desk. Although passively cooled for the most part, it does have a couple of fans in its graphics card which are just about audible when the room is otherwise silent. I wonder if the HF noise comes from these fans - maybe they ramped up during one of the takes? Easily checked - I'll take a few recordings of the ambient room noise.

 

If necessary, I can plug the M2 into my totally passive laptop for further mic captures.

 

That 19kHz noise may be inductor whine, move microphone to close to any switching power supply or DC-DC converter (PC, network switch, monitor display, graphics board, LED lamps etc) and see what device emit the noise :)

 

 

[yamamoto2002]

17 minutes ago, manisandher said:

 

The mic remained totally static for both captures. Why would there be more noise in one than the other, the second recorded literally seconds after the first finished?

 

Mani.

 

network traffic, background tasks, power state vary in time. computer display noise varies when drawing content is different.

 

 turn off room lights and computer monitor (set 1 minute timeout), and disconnect network, and use battery powered laptop. Insert several minutes of silence track before test signal and play

[yamamoto2002]

Here it is acoustic noise emitted from the power supply (?) of my sonically noisiest 27inch computer display (built-in speaker is muted)

 

AKG C414 → RME Babyface Pro → Asus UX31E laptop to record

 

Worst case (Very audible)

 

 

Best case (slightly audible from 1m distance)

 

 

Display switched off

[yamamoto2002]

2 hours ago, PeterSt said:

If you measure noises like this by microphone, the first thing what will happen is that you couple-in the electrical noise of what you think is audible. All the logic is there, ... except perhaps your perception in advance of what you should be able to hear. So Yamamoto, do you really hear this with your ears ?

 

Yes it is audible from day 1 when this monitor is arrived. It sometimes induces tinnitus. The perceived pitch is Si flat.

 

I read somewhere typical LCD monitor's power consumption is increased when dark object is displayed.

 

While this monitor is DisplayHDR 1000, it has powerful LED light source and a backlight dimming feature,  audible noise is increased when brighter object is displayed.

[yamamoto2002]

On 3/29/2021 at 11:35 PM, PeterSt said:

Redrawing of the screen under Windows (two pixels imply the full screen) implies 10000 (ten-thousand) times MORE activity than without GUI and drawing to the most native canvas. 10K times more.

 

I don't understand what you said fully but ... Windows OS has a InvalidateRect mechanism to redraw only small portion of the client area when small portion is changed.

 

Following image is a screen capture of `Dirty-Region Update Overlay' of Performance Profiling Tools for Windows Presentation Foundation.

 

During playback, PlayPcmWin redraws playback position slider (yellow region) on once in a 500 ms, and redraws playback time (purple region) once in one second. From the observation, Playback time update is reasonable. The slider position update causes slightly larger area redraw than absolute necessary, and it causes multiple dirty rect addition event :)

 

When mouse pointer hovers over the client area, more redraw event occurs.

 

If the app window is minimized, redraw event does not run. So it is possible to compare sonic impact of graphics redraw using Window minimizing functionality :)

 

Click wallpaper and press F5 to redraw full screen (if your keyboard is not customized)

 

Win+Ctrl+Shift+B to reset graphics driver

 

 

[yamamoto2002]

PlayPcmWin had had a ABX comparison function before but the function is dropped. Long ago foobar2000 did not have event mode exclusive WASAPI playback and USB asynchronous isochronous audio devices are not matured yet therefore playback stability can be vastly different to PPW (default is event mode), But now sonic difference is near-zero, use foobar2000 ABX comparator with WASAPI plugin.

 

Last I checked, foobar2000 does not create temporally file to play music, it can be checked with SysInternals Process monitor

[yamamoto2002]

42 minutes ago, March Audio said:

It also has an asio plugin if your dac supports it.

 

I don't recommend ASIO for music playback purpose, On Windows Xp or older versions, ASIO was the only option for bit-perfect playback. On Windows Vista SP1 (?) or later versions, there is equivalent functionality built-in to OS, called WASAPI exclusive mode and ASIO is not necessary anymore

 

For some reason, foobar2000 default is (ultimately) WASAPI shared mode and WASAPI shared mode is not recommended for serious testing, because it is difficult to set it up correctly

[yamamoto2002]

40 minutes ago, March Audio said:

Can you explain specifically what the issue is?  Asio works just fine here.  The RME utility indicates bit perfect and sounds fine.

 

Wasapi exclusive is indeed fine.

 

There are some unstable and buggy ASIO drivers there and they cause BSOD or other unstable behavior and WASAPI works fine with the same hardware, so IMO WASAPI is generally the better choice and recommended for music playback of audiophiles (for Windows 7 or later versions)

 

If ASIO works fine on your environment, there is no problem to use ASIO. On some audio devices, ASIO is only option for DSD 11.3MHz playback. On other professional recording devices (including RME), device sampling rate can be changed only via ASIO API

[yamamoto2002]

1 hour ago, March Audio said:

If anyone wants to listen to this original and allegedly "optimised" file you can find them here:

 

https://1drv.ms/u/s!AnQ0c7fb_4zLgk84MM9gWIU9Spme?e=eifCdJ

 

I think OneDrive uses data-deduplication to reduce their storage cost, your second upload is simply discarded and replaced with the link to the first uploaded data. even your first upload is discarded and replaced with someone uploaded data

 

http://www.ghassankarame.com/dedup.pdf

 

This is explanation pics of data dedup, those were originally created for sandy-k san 🙏

 

[yamamoto2002]

2 minutes ago, March Audio said:

Sorry Im lost.  Track A and track B both apear when I click the link.

 

two metadata  is created and both pointed to the same storage area

 

[yamamoto2002]

19 minutes ago, March Audio said:

 

 

Why dont you try the ABX test?  Lets see how reliably you can identify the "optimised" version when you dont know which is playing.

 

Sorry I cannot explain this complex mechanism more easily. Please look into the 3 pics above. On this storage, only unique cluster can be stored. Typically cluster size is 4096 bytes

[yamamoto2002]

5 hours ago, manueljenkin said:

Fyi foobar ab-x will copy the files to TEMP path hence it breaks the recommended usage of the tool.

Thank you for your confirmation work😀

[yamamoto2002]

2 hours ago, March Audio said:

This is a very generic/smplified description but the playback software will probably have a memory buffer for the audio data taken from the disk.   An asynch USB connection will ask for data as and when it needs it from the computer (ie when its memory buffer needs filling).  There is probably also more memory buffering in the USB driver.  The USB receiving end in the dac has a buffer.  Data is then accurately locally clocked out to the DAC chip.

 

This is about Windows 10

 

• Music player app may have a PCM buffer of arbitrary size on main memory.
• WASAPI exposes a PCM buffer of 3 ms to 3000 ms or so for music player app to write next PCM data to play.
• About WaveRT port driver: it has a cyclic buffer there. https://download.microsoft.com/download/9/c/5/9c5b2167-8017-4bae-9fde-d599bac8184a/WaveRTport.doc
• There is a software stack chart: https://docs.microsoft.com/en-us/windows-hardware/drivers/audio/usb-2-0-audio-drivers

  Quote

  

   

About USB Audio device:

• USB audio stream controller on USB Audio device has a cyclic buffer or ping pong buffer as a receive buffer of USB isochronous packets, then a FIFO buffer cascaded to send PCM to I2C bus of DAC chip.
• DAC chip itself has a FIFO buffer of 128 samples or so, in order to apply oversampling digital filter