Please forgive me if this is a dumb question. I am trying to figure out if the current standard for sampling rate is 44.1 kHz or 48 kHz. I have had some trouble with stations switching around between the two. CDs use 44.1 but, as I understand it, video uses 48. Going forward, which sampling rate is the one to use?
-
Get involved.
We want your input!
Apply for Membership and join the conversations about everything related to broadcasting.
After we receive your registration, a moderator will review it. After your registration is approved, you will be permitted to post.
If you use a disposable or false email address, your registration will be rejected.
After your membership is approved, please take a minute to tell us a little bit about yourself.
https://www.radiodiscussions.com/forums/introduce-yourself.1088/
Thanks in advance and have fun!
RadioDiscussions Administrators
44.1 OR 48?
- Thread starter bigdonho
- Start date
Both. 44.1 kHz is the norm for audio-only distribution. 48 kHz is the norm for audio used within video production.
In practice it doesn't matter what you use because all surrounding systems will re-sample to their preferred rate. Using Audition as an example of a common DAW (thought this is NOT a recommendation!), if you start a project at 48KHz, then import a track from a CD rip at 44.1, it will be resampled to 48KHz. In the PC OS, if the audio being set to an interface doesn't match the interface default, it gets resampled to whatever is appropropriate. Audio over IP systems may have a default you can pick, but if you're sending it analog it gets sampled at the default rate. If you have a direct digital interface on a PC, the OS will resample to the correct rate, or the AoIP system will.
Historically, of course the CD at 44.1 drove the digital audio market. Pre-CD digital recordings were often sampled at 50KHz using instrumentation recorders, but the CD system needed some way to record material to a digital format that could be edited, and the most practical and low cost means of recording digital audio was to format the data within a video field and use slightly modified video recorders, mostly 3/4" U-Matic machines. Sony pioneered the first affordable digital audio editor based on U-matic machines (DAE-1100). And their first PCM converter, the PCM-1600 from 1979, ran at 44.1 and 16 bits.
That frequency choice relates to 3 things:
1. Because of video bandwidth limits, the maximum number of 16 bit samples you can reliably in a single scan line is 3.
2. The maximum usable number of scan lines per field is 245 (in a 525 line frame there are 35 blanked lines and two fields, so 245 usable lines per field)
3. Assuming 60 fields per second for NTSC video, the formula is 245 x 60 x 3=44,100 Hz. Bingo, that's it.
And, you might have noticed a slight problem. When compatible color was introduced to NTSC video they nudged the B&W frame rate down from 30 frames/60 fields to 29.97 frames/59.94 fields which also nudges the sampling rate to 44.056KHz! So either the color video machine needed to be modified to run at 30/60, or it needed to be a special machine made just for PCM audio, OR...option 3...Sony introduced a Pro-Sumer PCM system that utilized consumer video recorders running at 29.97/59.94, and bumped the sampling rate down to 44.056. Which, by the way, could be easily warped up to 44.1 during editing with no discernable pitch bend if those recordings ended up on CD.
So the entire CD system was based on the video recorder-based PCM audio recording systems that slightly preceeded it. It all became Redbook standard, and off we go.
When digital audio for video arrived a few years later, we'd already learned that the analog anti-aliasing filter needed for 44.1 was a real pain, not easy to do well, and had negative impacts on some types of audio. Moving upward to 48khz (Nyquist at 24khz instead of 22.05) made that filter just a bit easier to make.
And then it all went wonky. We got over-sampling digital filters, so the finiky analog multi-pole beasts were no longer required. And 44.1 worked just as well as 48. Then people thought "more is better" and up went the sample rates. More is actually NOT better, and buys you nothing, but mythology prevails even today.
Really basic resampling of the early 1980s used to add 3dB of noise just to change from 44.1 to 48. But that's been figured out to, now it just works. And systems just do it.
The short version of this very long answer is, do what ever you need to do. Resample as few times as possible, but it doesn't really matter much at all. CDs and digital audio files like .mp3 are usually at 44.1, again, just because of CDs. Some automation play-out systems (iMedia for one) sticks to 44.1 for everything. Some digital transmitters and some codecs used for STL can to higher rates as needed, but most don't. Streams rarely....very rarely...to anything higher than 44.1.
You might keep in mind that mp3 has a built-in filter that cuts off everything above 15KHz with a pretty hard brick wall. That, 44.1 digital audio, and filtering for FM stereo (cut is between 15 and 17khz), might not be a great combo. But with cheap storage there's no need to keep audio files in mp3 now. Just use .wav at 44.1 and you're golden.
Historically, of course the CD at 44.1 drove the digital audio market. Pre-CD digital recordings were often sampled at 50KHz using instrumentation recorders, but the CD system needed some way to record material to a digital format that could be edited, and the most practical and low cost means of recording digital audio was to format the data within a video field and use slightly modified video recorders, mostly 3/4" U-Matic machines. Sony pioneered the first affordable digital audio editor based on U-matic machines (DAE-1100). And their first PCM converter, the PCM-1600 from 1979, ran at 44.1 and 16 bits.
That frequency choice relates to 3 things:
1. Because of video bandwidth limits, the maximum number of 16 bit samples you can reliably in a single scan line is 3.
2. The maximum usable number of scan lines per field is 245 (in a 525 line frame there are 35 blanked lines and two fields, so 245 usable lines per field)
3. Assuming 60 fields per second for NTSC video, the formula is 245 x 60 x 3=44,100 Hz. Bingo, that's it.
And, you might have noticed a slight problem. When compatible color was introduced to NTSC video they nudged the B&W frame rate down from 30 frames/60 fields to 29.97 frames/59.94 fields which also nudges the sampling rate to 44.056KHz! So either the color video machine needed to be modified to run at 30/60, or it needed to be a special machine made just for PCM audio, OR...option 3...Sony introduced a Pro-Sumer PCM system that utilized consumer video recorders running at 29.97/59.94, and bumped the sampling rate down to 44.056. Which, by the way, could be easily warped up to 44.1 during editing with no discernable pitch bend if those recordings ended up on CD.
So the entire CD system was based on the video recorder-based PCM audio recording systems that slightly preceeded it. It all became Redbook standard, and off we go.
When digital audio for video arrived a few years later, we'd already learned that the analog anti-aliasing filter needed for 44.1 was a real pain, not easy to do well, and had negative impacts on some types of audio. Moving upward to 48khz (Nyquist at 24khz instead of 22.05) made that filter just a bit easier to make.
And then it all went wonky. We got over-sampling digital filters, so the finiky analog multi-pole beasts were no longer required. And 44.1 worked just as well as 48. Then people thought "more is better" and up went the sample rates. More is actually NOT better, and buys you nothing, but mythology prevails even today.
Really basic resampling of the early 1980s used to add 3dB of noise just to change from 44.1 to 48. But that's been figured out to, now it just works. And systems just do it.
The short version of this very long answer is, do what ever you need to do. Resample as few times as possible, but it doesn't really matter much at all. CDs and digital audio files like .mp3 are usually at 44.1, again, just because of CDs. Some automation play-out systems (iMedia for one) sticks to 44.1 for everything. Some digital transmitters and some codecs used for STL can to higher rates as needed, but most don't. Streams rarely....very rarely...to anything higher than 44.1.
You might keep in mind that mp3 has a built-in filter that cuts off everything above 15KHz with a pretty hard brick wall. That, 44.1 digital audio, and filtering for FM stereo (cut is between 15 and 17khz), might not be a great combo. But with cheap storage there's no need to keep audio files in mp3 now. Just use .wav at 44.1 and you're golden.
The DV (Digital Video) standard that was introduced in 1995 and widely used in both consumer and professional gear gives a choice of either 2 channels of 16-bit, 48 kHz audio or 4 channels of 12-bit, 32 kHz audio.
I need to correct myself. I impled that digital audio for video drove the 48KHz sampling rate. Not quite true. 1982...the introduction of DASH recorders, all could do 44.1 and 48, with Sony and Studer adding bit depth to 24 bits. And 1987, DAT recorders ran at 32khz, 44.1 and 48kHz. I tend to forget DASH because it was high end recording studio only, and I tend to dismiss DAT because it was not as popular as the developers had hoped. Not bad, just not anything like recording to an HDD.
The DVD, 48khz 16 bit stereo was definitely part of the standard, but little used because Dolby AC3 (introduced for Dolby Digital sound on film in 1992) offered an efficient codec for 5.1 channels in less storage space than 2 channel 16/48. Nearly every film released on DVD used either AC-3 or DTS codecs, or both. They all ran at 48khz. I don't think I've ever seen a DVD with only 16/48 without another codec as well.
DVD-Audio (DVD-A, an audio only format on a DVD) standardized in 1999, but available in some form several years prior, allowed for a variety of channels, bit depths and sampling frequencies up to 192KHz stereo, and 5.1 up to 92KHz. The format was officially proclaimed dead in 2007, but some of those DVD-A discs I still have are flat-out spectacular.
Let's also clear one widely held misconception: sample rate/frequencie IS NOT resolution, it affects the highest possible audio frequency only. 16 bits at 44.1 is no greater "resolution" than 16 bits at 192KHz. It's just more data, and the higher sampling rate allows for a higher -3dB cutoff of the filters. Resolution is bid depth, number of bits, but that too is tilting at windmills because 24 bits has a theoretical 144 dynamic range that is unachievable in practice, and not reproducable in any way without endangering the health of the listener. 16 bits is more than enough for all listening. And that all makes 16/44.1 adequate, 16/48 more than enough.
PCM sampling is a "connect the dots" concept, not a "filter the stair steps" concept. There are no stair steps, never were, other than in some badly conceived graphics meant to explain PCM. They were and are wrong.
The DVD, 48khz 16 bit stereo was definitely part of the standard, but little used because Dolby AC3 (introduced for Dolby Digital sound on film in 1992) offered an efficient codec for 5.1 channels in less storage space than 2 channel 16/48. Nearly every film released on DVD used either AC-3 or DTS codecs, or both. They all ran at 48khz. I don't think I've ever seen a DVD with only 16/48 without another codec as well.
DVD-Audio (DVD-A, an audio only format on a DVD) standardized in 1999, but available in some form several years prior, allowed for a variety of channels, bit depths and sampling frequencies up to 192KHz stereo, and 5.1 up to 92KHz. The format was officially proclaimed dead in 2007, but some of those DVD-A discs I still have are flat-out spectacular.
Let's also clear one widely held misconception: sample rate/frequencie IS NOT resolution, it affects the highest possible audio frequency only. 16 bits at 44.1 is no greater "resolution" than 16 bits at 192KHz. It's just more data, and the higher sampling rate allows for a higher -3dB cutoff of the filters. Resolution is bid depth, number of bits, but that too is tilting at windmills because 24 bits has a theoretical 144 dynamic range that is unachievable in practice, and not reproducable in any way without endangering the health of the listener. 16 bits is more than enough for all listening. And that all makes 16/44.1 adequate, 16/48 more than enough.
PCM sampling is a "connect the dots" concept, not a "filter the stair steps" concept. There are no stair steps, never were, other than in some badly conceived graphics meant to explain PCM. They were and are wrong.
Thank you dc2bluelight. Excellent summary from you.