Subject:sampling and bit rate

Posted by: samcat

Date:1/27/2006 6:41:45 AM

I understand one can record at 44,100 sample rate along with 24 bit rate and dither down to 16 bit for cd playback. What I don't know is if you can record at a higher sample rate, say 48,100, 24 bit, and go down to 44,100, 16 for cd playback. As an aside, I recorded myself on guitar using 192,000, 24 bit and it sounded fantastic (the sound, not me) but of course the sound file is locked in my pc and can only be played back using my pc. Any thoughts on this without outfitting a professional studio?

Subject:RE: sampling and bit rate

Reply by: Chienworks

Date:1/27/2006 7:47:40 AM

Process / Resample.

Subject:RE: sampling and bit rate

Reply by: ForumAdmin

Date:1/27/2006 7:50:45 AM

Standard operating procedure is to track and edit at 24-bit and 48 or 96 kHz, then resample (Process->Resample) to 44.1 kHz, then dither and convert to 16-bit for CD delivery (for pros, this is the tail end of the mastering process).

Working at 192 kHz is going to eat up drive space in a hurry and I guarantee if you used some on-board SoundMax device, you weren't actually recording at 192 kHz. Most likely, Windows and the driver agreed on 44.1 kHz, and Windows upsampled it on it's way into Sound Forge. This doesn't really improve the quality of the audio coming out of the sound card as it has already been band-limited to half of its sample rate (probably 44.1 kHz).

You don't have to take my word for it and this is a good exercise even if you get hardware that supports higher sampling rates. Try recording the same thing at 44.1, 48, 96, and 192 kHz and see if you can spot the differences.

J.

Subject:RE: sampling and bit rate

Reply by: samcat

Date:1/27/2006 10:41:06 AM

By the numbers, 1 minute of audio:
at 192,000, 24 bit = 67.84MB, 11,577,600 samples
at 44,100, 16 bit = 10.39MB, 2,659,230 samples

Subject:RE: sampling and bit rate

Reply by: ForumAdmin

Date:1/27/2006 1:35:46 PM

If you mean this as evidence that you are recording at 192 kHz, let me re-phrase, as this comes up often:

I very much expect that the A/D in this sound card is delivering 44.1 kHz and the Windows kmixer component is "being nice" and upsampling to 192 kHz since that is what you are asking for from Sound Forge. Yes, you're getting 192 kHz audio in Sound Forge, but the actual frequency content is still bandlimited to 22.05 kHz (half of 44.1 kHz, the A/D's actual sample rate).

Playing back a 192 kHz file? Yup, the kmixer is downsampling to 44.1 kHz before delivering it to the sound card for D/A.

Moral of the story? The Microsoft kmixer exists to allow simultaneous system-wide access to the audio device, but unnecessary conversions should be avoided in production. Know what your hardware physically supports and use it's maximum bit-depth and sample rate (typically 24/96 or 24/192 for pro cards, and typically 16/44.1 or 16/48 for consumer and on-board devices).

Other driver models (like ASIO or kernel-mode streaming) bypass the kmixer and do the "right" thing by disallowing unsupported formats altogether.

J.

Subject:RE: sampling and bit rate

Reply by: samcat

Date:1/28/2006 3:07:48 PM

Not so sure. If you try to burn a c.d. in other than 44,100/16 your get an error "unrecognized format" which leads me to believe the sampling rate/bit rate are indeed at the selected value.

Subject:RE: sampling and bit rate

Reply by: Chienworks

Date:1/28/2006 4:49:00 PM

samcat, if you read the moderator's replies carefully, you'll see that "J." is telling you that your file probably is 192K 24bit. However, the data in the file came from a 44.1K source. What this means is that the recording process duplicated the samples from the 44.1K source to filll in the extra samples. Since 192K / 44.1K = 4.354, what's happening is that each sample from the source is copied 4 or 5 times in the file you've recorded. Since the data is duplicated the actual data resolution in the file is still 44.1K, even if the samples are at 192K. Likewise with the bit depth, the source is probably 16 bit and the recording process is flushing out the other 8 bits with zeros to fill the full 24 bits in the file. The extra 8 bits contain no information so once again the data resolution is 16 bits with the other bits being wasted.

Subject:RE: sampling and bit rate

Reply by: samcat

Date:1/29/2006 12:19:42 PM

I'm using a sound blaster extigy rated at 24 bit, 96 Mhz so I assume I am getting a true resolution at that level. What I don't understand is why the recording process duplicates the samples to allow this to happen. If I try to connect a soundcard rated at 44.1/16 bit to software set for 24 bit, I will get an error message. Usually if you try to do something "illegal" the software and or hardware program will balk.

Subject:RE: sampling and bit rate

Reply by: Geoff_Wood

Date:1/29/2006 12:59:09 PM

I would add that resampling artifacts will degrade a final 44k1 piece originally recorded at 48K more than doing it all at 44k1 in the first place.

If you are recording at 96k or whatever, then you are doing that for some specific perceived reason, so just go for it. But 48K to 44k1 - forget it !

geoff

Subject:RE: sampling and bit rate

Reply by: samcat

Date:1/30/2006 4:19:43 AM

From "Home Recording, June 2002, Terms, Trends & Technologies in Digital Recording by Jon Chappell"..."Most people agree that recording at 24/48 (and higher) and then dithering and sample-rate converting to 16/44.1 sounds better than initially recording at 16/44.1."

Subject:RE: sampling and bit rate

Reply by: MarkWWWW

Date:1/30/2006 5:33:26 AM

If I recall correctly the Extigy was one of the group of Creative soundcards that were falsely advertised as supporting 24/96 but in reality did not do so.

There was a legal action against Creative who agreed to compensate users in some way. I'm not sure of the details but if you can understand legalese you can see them at http://www.audiocardsettlement.com

The upshot is that you are probably not getting true 24/96 with that soundcard.

Mark

Subject:RE: sampling and bit rate

Reply by: samcat

Date:1/30/2006 9:18:37 AM

great, one more reason to avoid soundblaster. thanks for all the help.