Jump to content


  • Content Count

  • Joined

  • Last visited

About yamamoto2002

  • Rank
    Freshman Member

Personal Information

  • Location

Recent Profile Visitors

3547 profile views
  1. On locally stored video file playback, if PC sends video and audio using HDMI, its image sequence and audio will be sent completely synchronized manner. also HDMI automatic lip sync function may work. If video is sent to HDMI and audio is sent to asynchronous USB DAC, there are two free-running clocks in the system: GPU RAMDAC clock and USB clock, and two clocks does not synchronized at all. In this situation, which clock becomes master is depend to video playback app implementation. I checked Microsoft media player on Windows 7 and found async USB DAC clock becomes master of movie file playback and audio data sent to USB DAC is bit-perfect while video frame may be skipped intermittently. Another app implementation possibility is, audio data is filtered using asynchronous sample rate converter (such as IAudioClockAdjustment) to match to video clock timings, on this route, all video frame is sent correctly but audio data becomes not bit-perfect.
  2. About 20 years ago of issue of local computer magazine 月刊アスキー, there was an article about sound difference of identical DAW project file stored on different hard drive. One drive is IDE HDD and another drive is SCSI HDD, and a music composer says SCSI HDD sounds better. On the next issue, the cause of sound difference is found, it is caused by sound stutter due to slow drive transfer speed of IDE HDD
  3. This sure reduce the chance of hash collision, two larger than 512 bit files may happen to share the same MD5 and SHA256 and MD4 checksum value approx in of probability.
  4. Identical checksum does not necessarily mean two files are the same bit by bit. For example, MD5 checksum is 128bit fixed length integer value, therefore two larger than 128 bit files may share the same checksum value. Chances are low (slightly higher chance than of probability) The following example files are created deliberately. WAV files a.wav and b.wav in attached SameMD5Wav.zip share the same MD5 checksum value 161b403cd39f1f3f6cffb64b4dd8ccb9 but Waveform is different, beginning part difference is obviously seen. C:\tmp>CertUtil -hashfile a.wav MD5 MD5 hash of a.wav: 161b403cd39f1f3f6cffb64b4dd8ccb9 C:\tmp>CertUtil -hashfile b.wav MD5 MD5 hash of b.wav: 161b403cd39f1f3f6cffb64b4dd8ccb9 SameMD5Wav.zip
  5. I have this kind of experience. It is super weird, people hear 'sound difference' of every released version, even with the corner case bug fix that code runs only on metadata processing (it runs on file header loading, before playback starts) fix, that bug causes app crash when conditions are met. Then, I wrote changelog what is changed on every release and put source code change history link, and this kind of sound difference email is reduced significantly. It seems, people hear sound difference when new version is released without enough information of the revision changes, for people to judge the change causes sound difference or not, and people judge reasonably when enough information is available. Also it seems people do read app changelog more carefully than I thought!
  6. Audio tape recorders use AC bias to improve signal linearity. AC bias signal frequency effectively defines upper limit of audio tape recorders of recording frequency , it is typically configured 50kHz in R2R
  7. I used 10 USB cd drives connected to 2 PC to rip 2704 Audio CDs. USB CD drive was cheap, $15 each. It took 4 days This is a screenshot while testing how many USB DVD drives are too many for one PC
  8. Self-propelled floor-standing speaker, it returns their home position when it is not used like Roomba
  9. On 1995, my computer had 16MB memory (memory costs $1200) and 420MB harddrive. I read an audio magazine article about a CD transport with digital out, it says : "Servo motor to move optical pick-up draws relatively large current and when it actuates, circuit power voltage temporally drops, this causes small temporal error of output digital signal, it is called jitter" and I thought "Then store all the CD transport output data to the 650MB (Audio CD data size) memory buffer and when all the data is extracted from CD, turn off the CD transport and start sending data to DAC with pulling data from the memory buffer with clean clock timings without jitter caused by servo motor."
  10. Typically amplifier input is high impedance input, that means the input circuit of amplifier senses voltage, it draws very small amount of current. Therefore output power of the sound source device is not relevant, output voltage is important parameter. I read specification of R6 Pro and found its analog line out is 2.0V rms, while its single-ended headphone output terminal can provide 2.8V rms when headphone volume is max. Maybe first thing to try is to connect R6 Pro headphone out (instead of line out) to your amp. If it does not provide enough sound pressure, get analog pre-amp to increase signal voltage
  11. I think 10 to 30 years old Goldmund Alize DACs do not have digital oversampling filter. Instead, it has super steep brickwall analog low pass filter of 20 stage or something to cut all noise (aliasing noise which is inevitably generated by non-oversampling digital circuit) above 20kHz. IIRC there is an analog all pass filter circuit cascaded in order to compensate phase distortion caused by analog brickwall filter. It seems output analog signal is clean, similar to the conventional oversampling dac.
  12. WASAPI shared mode do nothing to PCM data when There is only one audio session coming into the mixer Original PCM bit depth is 25bit or smaller (theoretically, I verified up to 24bit using S/PDIF digital out and digital PCM recorder) All Sound enhancements are turned off Session volume and endpoint volume are both 100% Original PCM magnitude does not exceed 98.5% of full swing Shared mode sample format bit depth is set to 24bit or 32bit (not 16bit, where dither is applied) Even bit perfect playback (original PCM values are outputted unaltered) is possible in WASAPI shared mode when those condition is met. Limiter APO temporally reduces gain when incoming signal exceeds 98.5% but some times after incoming signal becomes smaller 98.5%, gain returns 1 (I don't remember time scale but it is smaller than 1second) Sound distortion occurs when Limiter APO is activated and its distortion level is up to 0.13dB, some people in Japan reports they noticed sound difference compared to original signal but they said difference is subtle. I cannot hear the difference Resampling artifact may become larger problem than mixer (Limiter APO) distortion, it is important to adjust shared mode sample rate to match to original PCM sample rate on shared mode playback for critical listening (this is true when resampling algorithm is poor). I don't know implementation details but limiter APO behavior should be the same across all type of CPUs CPU difference do cause sound difference when CPU computing performance is relatively low and other tasks run while playback: other tasks take CPU time and playback thread fails to deliver PCM signal on time: this causes sound stuttering like vinyl needle jump. This is always very obviously heard as annoying playback failure when it happens. This happens more frequent in Windows 2000 era and Windows Vista era of computer and things gradually improved
  13. I'm curious what Tsimané people do when a man and a woman sing the same song in unison 🤔💭
  14. Windows 10 privides several different sample rate converters to resample PCM signal for apps, with different conversion quality. Also apps may have own sample rate converter. I don't know Amazon HD uses which API to play sound, but SRC can be bypassed when source sample rate and shared mode sample rate is matched. Shared mode's sound altering filters can be disabled on Enhancement tab of Sound Control Panel (Fig. 2). When SRC is bypassed, remaining problem of altering sound of Windows shared mode is Limiter APO, which reduces gain temporary when incoming PCM is too hot. It starts to work when incoming sound magnitude exceeds 98.5% of full swing. Windows provides these sample rate converters: MME Resampler: Used when App play sound using old MME API. I don't check its conversion quality of Windows 10 versions. Direct Sound Resampler: Short-delay (not linear-phase) resampler. Conversion quality is low. Media Foundation Resampler: Latest converter of Windows. Linear-phase. Conversion quality can be controlled by app. App's own resampler: Music player apps may have their own resamplers for their needs. I believe all SRCs of Windows 10 provides use CPU. Most app's own SRC use CPU, small number of apps use GPU to perform SRC. Also there are audio hardwares such as Lynx AES16e-SRC that contain hardware sample rate converter chip. Fig.1 Fig.2 Enhancement tab of Sound Control Panel
  • Create New...