logo Sign In

Please explain "Bit-Perfect Audio Capture" for LaserDisc PCM

Author
Time

i'm a bit confused because to my knowledge capturing the pcm audio of a laserdisc through optical (toslink) is bit-perfect. i mean what else can one do? increasing the capture from 16bit to 24bit wont do anything, except a higher filesize. increasing the samplerate from 44.1khz to 48khz on capture wont do anything either, in terms of quality.

i mean, the laserdisc pcm is 16bit @ 44.1khz - what more can you do?

so what do people refer to when they talk about bit-perfect and saying that a normal digital capture wont be enough?

please enlighten me :)

Author
Time
 (Edited)

A "bit perfect" audio capture means that Every. Single. Bit. in the capture is the same as what's stored on the digital medium (in this case, the digital track on an LD).

Some older sound cards did digital post-processing (programmed in the driver) on the TOSLINK input (e.g., band-pass filtering) without notifying the user and without giving the user a way to disable it. Hell, some even did it in the hardware.

I know this sounds absurd since all the sound card needs to do is dump the PCM bytes it receives over the TOSLINK input onto the HDD, but this wasn't (and isn't) always the case. This is why some sound cards advertise "bit perfect" audio capturing.

It's not a gimmick. It's a response to the stupidly-written overzealous drivers of yesteryear.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

thanks antcufaalb :)

is there a way to test this?

Author
Time
 (Edited)

Also, here's another great explanation of "bit perfect": https://code.google.com/p/cmediadrivers/wiki/Bitperfect

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

AntcuFaalb said:


http://www.head-fi.org/t/463089/is-your-s-pdif-bit-perfect-heres-a-test
Does bit-perfect output prove anything about the input path?
I won't leap to that conlcusion.


zeropc said:


is there a way to test this?
Record something twice, thrice, or (n+1) times.
Compare them.
Do they match?

If they've been resampled, then they almost certainly won't match.
Zillion-to-one probability

However, in practice you must take into account the “fuckwit factor”. Just talk to Darth Mallwalker…
-Moth3r

Author
Time

A simpler way would be to do two captures of the same source at significantly different recording levels - if one recording is a higher volume than the other, they're being resampled (since gain cannot apply to a pure digital transfer).

Since you'd naturally expect to be able to adjust the gain, the default setting on your hardware will almost certainly be set to resample - sound cards with bit-matching (such as the Creative X-Fi mentioned above) have a specific option than needs to be checked to enable it.

Author
Time

Simpler? That's easy for you to say!
My M-Audio card don't have input level controls.

However, in practice you must take into account the “fuckwit factor”. Just talk to Darth Mallwalker…
-Moth3r

Author
Time

so i did a few testings, like you guys suggested.

i'm on a mac btw ;)

playback testing

1. i tested a dts-cd wav in vlc (latest version) and it plays fine through system speakers

2. the same dts-cd wav in vlc through toslink was a no go. the system settings simply switched the output device for an unknown reason

3. muxed the dts-cd wav to .mka and then vlc played it just fine through toslink and my amplifer saw it as dts and playback was fine

4. same mka file in mplayerx gave me white noise - could be because of an older version i'm using, as the newer version gave me serious playback problems on some movies

digital capture testing

1. i cannot change the gain level when i plug in my optical cable. adobe audition would always jump back to 0db

2. multiple capture of the same file - 3 to be precise and they all looked the same. there was no change in the waveforms hight or what not...

conclusion

i guess the mac is doing bit-perfect capture and output (if used in certain container formats like .mkv/.mka). maybe i try to get myself an external capture device. would the M-Audio Firewire 1814 a good choice?

Author
Time

I'm not aware of any problems with Mac doing resampling of SPDIF capture. Just idiotic PC hardware makers.

Why do you want to spend money when you've already found a working solution?

Author
Time

zeropc said:


2. multiple capture of the same file - 3 to be precise and they all looked the same. there was no change in the waveforms hight or what not...

Looked the same?!?
Your eyes can deceive you-- don't trust them.

Let the computer compare the data byte-for-byte.
Computers are good at that sort of thing.

Windows ships with FC.EXE (File Compare) while UN*X derivatives have their cmp (CoMPare).
OSX should have cmp command, yes?

Exact Audio Copy (EAC) offers a graphical interface to compare 16/44.1 WAVs sample-for-sample.

Cross-platform shntool is my personal favorite
.

However, in practice you must take into account the “fuckwit factor”. Just talk to Darth Mallwalker…
-Moth3r

Author
Time

i've downloaded xACT, which comes with shntool and i get the following result on a testfile

shntool [cmp]: warning: WAVE data size differs between these files -- will check up to smaller size
Comparing [/Users/zeropc/Desktop/wav-compare/Audio 2.wav] (0:17.867) and [/Users/zeropc/Desktop/wav-compare/Audio.wav] (0:15.963) :   0%  18% ERROR
shntool [cmp]: error: WAVE data differs at byte offset: 466847

is that okay or bad?

Author
Time

Might be a false alarm.

When you made those recordings, did you press Record button before pressing Play?
If so, then you'd have some digital silence at the start of each WAV, and likely not the same amount each time.
That can easliy confuse shntool.
Shntool takes the first sample from the first file and compares it against the first sample from the second file.
They match-- both samples have a value of Zero (digital silence).
Likewise the 2nd samples from each file match each other,
and the respective 3rd samples match, etc. . . . because they're all Zeroes.
shntool would mistakenly conclude that the two WAVs are aligned, when they're probably not.
Later when it reaches 18% one file might contain Non-Zero data
while the other file is still silent zeroes. Hence the error.

Try trimming the digital silence from the start of each file first.

Actually shntool will presume the two files are already aligned, unless you tell it otherwise by using the "-s" option.
Then shntool would try to detect how much time-shift between files.
Not sure if xACT will pass that option to the underlying shntool, or not.

You could run shntool yourself from a command prompt, such as
shntool cmp -s Take1.wav Take2.wav
In that case the files needn't be perfectly aligned, but within three seconds (default setting).
Still might need to trim leading zeroes though to avoid a false positive as above.

If using xACT (presumably without the -s option?) then you'd have to align the files yourself before calling shntool.

However, in practice you must take into account the “fuckwit factor”. Just talk to Darth Mallwalker…
-Moth3r

Author
Time
 (Edited)

Darth Mallwalker said:

When you made those recordings, did you press Record button before pressing Play?

no, there was enough time before the audio started. but usually i would press record before pressing play.

anyway... i tried to use the -s cmd with synced test-files and i get another weird result

/bitperfect_test/shntool cmp /bitperfect_test/ww1_mixdown.wav /bitperfect_test/ww2_mixdown.wav
Comparing [/bitperfect_test/ww1_mixdown.wav] (0:14.21) and [/bitperfect_test/ww2_mixdown.wav] (0:14.21) :  21% ERROR
shntool [cmp]: error: WAVE data differs at byte offset: 1

when i use the same files without the -s cmd, then i get this

/bitperfect_test/shntool cmp -s /bitperfect_test/ww1_mixdown.wav /bitperfect_test/ww2_mixdown.wav
Scanning [/bitperfect_test/ww1_mixdown.wav] and [/bitperfect_test/ww2_mixdown.wav] :   0% ERROR
shntool [cmp]: error: these files do not share identical data within the first 529200 bytes.

Author
Time

^Yeah, that's weird.

Maybe it's time for sanity checks.
Let's begin with a raw recording, and no leading silence.
Press Play, wait until the music starts before pressing Record.
I'll call it Take3.wav

Should match itself (we hope!)
shntool cmp Take3.wav Take3.wav


Next we'll test if "-s" is working correctly.

echo 0:02 | shntool split Take3.wav
That should produce two output files:
split-track01.wav (should be exactly 2 seconds long, or 88.200 samples)
split-track02.wav (the rest of it)

shntool cmp Take3.wav split-track01.wav
shntool cmp -s Take3.wav split-track02.wav
shntool cmp -s split-track02.wav Take3.wav


Are the results as expected?

However, in practice you must take into account the “fuckwit factor”. Just talk to Darth Mallwalker…
-Moth3r

Author
Time

AntcuFaalb said:

Also, here's another great explanation of "bit perfect": https://code.google.com/p/cmediadrivers/wiki/Bitperfect

I have one of those compatible sound cards, and I needed to install those drivers in order to get AC3/DTS SPDIF pass though.

I haven't tested SPDIF input, though.

Guidelines for post content and general behaviour: read announcement here

Max. allowable image sizes in signatures: reminder here

Author
Time

zeropc said:

i’m a bit confused because to my knowledge capturing the pcm audio of a laserdisc through optical (toslink) is bit-perfect. i mean what else can one do? increasing the capture from 16bit to 24bit wont do anything, except a higher filesize. increasing the samplerate from 44.1khz to 48khz on capture wont do anything either, in terms of quality.

i mean, the laserdisc pcm is 16bit @ 44.1khz - what more can you do?

so what do people refer to when they talk about bit-perfect and saying that a normal digital capture wont be enough?

please enlighten me 😃

Forgive the lateness of this post, I thought it may be of use to anyone still looking into this.

My understanding is Laserdisc PCM is 16bit/44.1kHz - no more. So whatever capture you do, it would be advised to keep it 44.1kHz as record at 48kHz (or higher) will create sync issues with the video.

Yes, you could set the DAW at 24bit/44.1kHz, but the only advantage of that would be if you need to use plug-in to alter the sound.

I’ve never tried to capture from toslink to DAW as my LD Player doesn’t have an optical out for digital stereo tracks. If it’s just PCM, then I would assume any DAW interface with a toslink input should receive the digital stereo signal.

DTS 5.1 audio, while coming from the same toslink output, needs a decoder to turn it form code into 5.1 channels of audio. This is normally handled by an AV receiver amp. Same deal as the RF AC3 tracks that need a demodulation to turn it into a 5.1 digital signal that would go to an AV receiver amp via toslink or coaxial. So other than using a line level convertor to convert your AV receivers speaker outputs into line level to input to an analogue 6 channel input - I couldn’t tell you how to capture an un-decoded AC3 or DTS signal to 6 channel digital PCM audio while maintaining digital all the way down the chain.