RIP track from CD with Windows Media Player (ver 12)
Import the track. (drag and drop or file import audio…)
Export the track. (as wav signed 16-bit PCM)
Comparison (kdiff) shows the original and the exported are not the same.
I’m aware of the differences introduced by rounding and dither when converting to a lower bit depth. But my understanding is that CD RIP WAV should be 16 bit PCM. Same as the export. So where does the difference come from?
I realize the difference is probably not even audible. The interest here is for understanding what is taking place.
According to VLC media player both the original and exported files are:
Codec: PCM S16 LE (s16|)
Type: Audio
Channels: Stereo
Sample rate: 44100 Hz
Bits per sample: 16
Reimporting and exporting the exported file does not introduce a difference. The difference is only when exporting the original CD RIP wav file.
Also, the same import export process using VLC Media Player to get the original CD RIP WAV results in binary identical exported file.
Audacity works internally in 32-bit float, so export is always coming from 32-bit float to whatever you select as the Export format.
Ideally, Audacity would see that the samples are all exact 16-bit values, and automatically disable dither, but as yet Audacity has no mechanism to do that (this is logged on the developer’s “To Do” list).
In order to get exact 16-bit sample values, you need to temporarily disable dither in “Edit menu > Preferences > Quality”.
Right. Dither is disabled and both the source (CD RIP) and destination (export) are 16 bit and no processing is being done before exporting. So there should be no rounding or dithering. Proved this out with a RIP of the same track using VLC Media Player. The original and the exported files are bit exact. As is expected. So why not with the Windows Media Player RIP? What’s different?
“If dither is not disabled in Audacity, Audacity will add dither, even if it is not actually needed.”
Realized that after posted and edited it. But nonetheless it is disabled.
“If you compare the ripped WAV file from WMP with the ripped WAV file from VLC, are they the same or different?”
No they are considerably different. VLC RIP is about 0.4% smaller. About 30KB of 7.45MB.
I’d GUESS the difference is metadata since KDiff shows ASCII/text “WAVELISTINFORART” in one of the files.
Exact Audio Copy, CueRipper, and a few others support [u]AccurateRip[/u] which will compare your rip to other rips in it’s database. (And, it won’t be fooled by metadata or any non-audio information.)
May the difference be dued ton a offset in the timing?
This reminds me when I tried to compare WMA with MP3 by encoding an audio CD in lossless WAV, WMA (192 kbps), and MP3 (320 kbps), all them with Windows Media Player; then I imported them into audacity, inverted the phases of the lossless WAV, and played tracks by two at a time, expecting to hear something near to silence. Didn’t work because the waves of each secuency where not aligned; and aligning them manually became almost impossible.
The first 255 bytes of WMP rip wav has various differences compared to the first 39 bytes of VLC rip wav. Guessing this is header and metadata differences.
The last 32928 bytes of WMP rip wav are additional compared to VLC rip wav. Guessing based on size that this may be a default embedded image. Perhaps that music note image WMP displays. But just a wild guess that is more likely wrong than right.
Everything in between is bit exact.
WMP rip wav vs. Audacity exported wav
The first 255 bytes of WMP rip wav has various differences compared to the first 39 bytes of Audacity exported wav. Looks like the differences have to do with the WAVELIST INFOIART and other metadata.
Last 279 bytes of Audacity exported wav are additional compared to the WMP rip wav. The LIST, INFOINAM metadata appear to be here instead of at beginning.
