Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 'non-standard' sample rate (Read 4007 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

'non-standard' sample rate

I have a question concerning the sample rate of ADC in sound cards. Let's say it can work up to 48 kHz, my question is can you actually record at any rate up to this limit, for example use values like 47156 Hz, or should you always use 'standard' ones like 48 kHz, 44.1 kHz, 32 kHz, etc.?
I have read in various topics on this forum that some people used to change the sampling rate, for example from 44100 Hz to 43235 Hz to correct pitch of a turntable which went slightly off-speed and then resampled the recording back to 44100 Hz, or even recorded 45rpm vinyl at 33rpm, changed the rate appropriately to 59535 Hz and again resampled to 44100 Hz. (There was a reason that the turntable suffered from an audible wow/flutter at 45rpm if I remember correctly, though the RIAA curve got slightly off in that case.)
When I thought about it, I came to an idea, that if a turntable runs e.g. 1 % faster, it could be recorded at 44541 Hz and then the sample rate changed to 44100, and avoid any later resampling hassle. Certain applications, e.g. GoldWave, allow you to type any rate you like to record at. However, I am not sure if the job is done by the sound card itself, or if it uses any closest 'standard' sample rate value which would cover the frequency range you wish to record, which would be 48000 Hz in our case and then the used recording application make some kind of quick real-time resampling to the mentioned 44541 Hz, which might not be of that good quality as if you did it later yourself in an application like SOX.
I am aware that various applications offer different kinds of time-stretching effects, but when I tried these, it always resulted in audibly worse recording than that with changed sample rate.
I hope someone will be able to answer my question.

(I am not quite sure whether I put my question in the right section or not.)

'non-standard' sample rate

Reply #1
Your ADC can run at any arbitrary rate but you won't have access to the pll circuits needed to generate nonstandard rates without designing your own device or buying a special unit with an external clock input.

'non-standard' sample rate

Reply #2
I don't see any reason to record at non-standard sample rates.  And, I'm pretty sure any oddball sample rates are created in software after A/D sampling.  The (true-original) sample rate is derived in hardware from a high frequency (MHz) crystal.

Quote
I am aware that various applications offer different kinds of time-stretching effects, but when I tried these, it always resulted in audibly worse recording than that with changed sample rate.
You shouldn't be hearing artifacts.  Sample rate conversions should be transparent (as long as you stay above 44.1kHz or so).  You can get audible artifacts when you change speed or pitch independently, but hopefully that's not what you are doing. 

Quote
...to correct pitch of a turntable which went slightly off-speed
I've NEVER had a speed problem with ANY properly functioning turntable.  Every turntable I've owned either has a synchronous motor (locked to the line-frequency) or adjustable speed and a strobe.    The line frequency is normally very good and if you live in a country where the frequency is unstable, it's likely to be unpredictable so you won't be able to compensate by tweaking the sample rate in advance.

I have heard of rare cases where the pitch of a record was off (usually in older recordings caused by mis-tuned instruments).    That's not something you'd notice unless you have perfect pitch or unless you are a musician trying to play-along with the record.    Although it's not a speed problem it can be corrected by adjusting the speed, and the very-slight tempo change is usually not noticeable. 

Quote
(There was a reason that the turntable suffered from an audible wow/flutter at 45rpm if I remember correctly
I've NEVER heard wow or flutter from ANY properly functioning turntable.  Again, the most common cause of wow (besides a defective turntable) is a defective record with an off-center hole.

Quote
if I remember correctly, though the RIAA curve got slightly off in that case.)
Correct.


P.S.
BIG PICTURE -
It's analog vinyl...  The weak links are all on the analog side.    Unless you have a noisy soundcard I wouldn't worry about the digital part of the recording process.    When it comes to cleaning-up the "snap", "crackle", and "pop", then you can get picky again.

'non-standard' sample rate

Reply #3
Quote
I don't see any reason to record at non-standard sample rates.  And, I'm pretty sure any oddball sample rates are created in software after A/D sampling.

That's what I thought but I was not sure. So basically it's better to use always 44.1 or 48 kHz.

Quote
You shouldn't be hearing artifacts.  Sample rate conversions should be transparent (as long as you stay above 44.1kHz or so).  You can get audible artifacts when you change speed or pitch independently, but hopefully that's not what you are doing.

When I change the sample rate and then resample, it sounds perfect, similar as if I changed the speed on a turntable. No audible problem with resampling really. But if I use the effect called time-streching (where you can type the wished final length or percentage how much you want to make the recording slower/faster) the sound becomes somehow dull, not terrible, but definitely quite an audible difference. Of course, I am not changing the pitch or speed only without affecting the other.

Quote
I've NEVER had a speed problem with ANY properly functioning turntable.

I have noticed that many old ones have this problem. Sometimes it is really noticeable, especially if it goes 2% faster and more. In my case it's not audible at all, but it's measurable. I tested it on tracks which I have on vinyl and CD's as well, I cut the track exactly at the same point of a beat at both ends and compared to each other, the result is a constant number, it makes 0.40% faster for records played at 33 rpm, on the other hand records played at 45 rpm run a bit slower than they should. If that number was not constant, I would attribute that difference to very light inaudible wow/flutter. Personally, I would left it alone if there wasn't an option to fix it without any artifacts, especially when in my case the speed difference is inaudible. However, since I have such an option why not to fix that while I am at it.

Quote
It's analog vinyl...  The weak links are all on the analog side.    Unless you have a noisy soundcard I wouldn't worry about the digital part of the recording process.    When it comes to cleaning-up the "snap", "crackle", and "pop", then you can get picky again.

I do realise that, but actually why make things worse (in a technical sense here, though inaudibly) than they need to be.

Cheers.

'non-standard' sample rate

Reply #4
Pitch correction algorithms have variable quality.  Try a better one?

'non-standard' sample rate

Reply #5
Keep the two different tools from each other (and I have no idea what they are called in the various applications):
- just change speed. That changes pitch and the length accordingly. Like moving to 45rpm.
- change pitch without changing speed. That is a more involved process where new samples are inserted if you tune up. (Then there are those algorithms that change speed without changing pitch.)

If a vinyl rip has wrong pitch - be it the playback or the record - then it is due to wrong speed. You want the former of the above two - good news, as that is the simplest to perform.


As for recording "at different sample rate" in order to compensate: my educated guess would be that such an option means that your software does resample, and that you might just as well use a rate native to your soundcard. 48 kHz should be enough - you don't change pitch by 20 percent, do you? - but if supported, it is costless to record at 96/24 and then reduce file size in the end, and it might reduce the risks of clipping and all the sorts of things that could make you want to rip once again.