[GreenBow]

If you start with a 192KHz sound file, can you easily convert it to 44.1KHz sound file. Or would it be only possible to 48KHz as a multiple of 192KHz?
 
Secondly can you convert from 24bit to 16bit.
 
Meaning overall, could you covert/compress a 24bit 192KHz audio file to a 16bit 44.1KHz (RedBook) audio track. Would this result in a file that would be the same as the track having been recorded in 16bit 44.1KHz initially.
 
Sorry if this has been covered extensively as I guess it's for sure been asked before.
 
Yours thank you, GreenBow.

 

[spruce music]

You can convert to any sample rate. You can convert to 16 from 24 using dither is better. The result may not be precisely like if it was 16/44 originally but close enough.

 

[VNandor]

Yes, it can be done. They wouldn't be the same in a way they digitally nulled out each other out. However they would virtually be the same for playback purposes.

 

[GreenBow]

OK. That's a good result. Thank you.
 
Please is there an easy way to do it that anyone knows of/ I am hoping to take something like 24bit 192KHz  audio track, and be able to convert it to 44.1KHz. Then end up with a physical audio track in 44.1KHz from the original HD-track.
 
(I don't simply mean playing an HD-track through a media centre at 44.1KHz. I mean actually ending up with another physical track the same, at another sampling frequency, being RedBook.)  
 
Are there specific software applications for it, or would you do it in your Digital Audio Workstation?

 

[RRod]

  OK. That's a good result. Thank you.
 
Please is there an easy way to do it that anyone knows of/ I am hoping to take something like 24bit 192KHz  audio track, and be able to convert it to 44.1KHz. Then end up with a physical audio track in 44.1KHz from the original HD-track.
 
(I don't simply mean playing an HD-track through a media centre at 44.1KHz. I mean actually ending up with another physical track the same, at another sampling frequency, being RedBook.)  
 
Are there specific software applications for it, or would you do it in your Digital Audio Workstation?

 
Audacity and SoX would both be ample. The incantation in SoX would be:
sox in.flac -b 16 out.flac rate -v 44100 dither -s
 
Re resampling to a non-integer multiple:
You can see this as a rational resampling, first upsampling 147x and then downsampling 640x. That is:
192000*147/640 = 44100

 

[GreenBow]

   
Audacity and SoX would both be ample. The incantation in SoX would be:
sox in.flac -b 16 out.flac rate -v 44100 dither -s
 
Re resampling to a non-integer multiple:
You can see this as a rational resampling, first upsampling 147x and then downsampling 640x. That is:
192000*147/640 = 44100

 
Great. Informative. That's what I was hoping for. I have Audacity, but I am currently looking at DAWs. I am going to run both Cubase and Pro-Tools trail versions in the coming weeks. (Probably stating this week. I am wrangling the trial downloads around my current ISP cap.)
 
I used to own Cubase, and used both DAWs at university. That was good few years ago now though, and I don't if we had such high sampling rates then. I think I recall 48K.
 
I am jumping back on the horse so to speak, and getting back on with my hobby. Thank you for helping. As long as I know it can be done, and easily, then I am OK.
 
I can't help thinking that doing direct multiples is better though. That way I guess the compression literally takes, e.g. the fourth sample in converting 192KHz to 48KhHz.
 
Anything more complicated like you explained ~ 192000*147/640 = 44100, would mean extrapolation I guess. Then it comes down to quality of DAW, I guess.

 

[RRod]

   
I can't help thinking that doing direct multiples is better though. That way I guess the compression literally takes, e.g. the fourth sample in converting 192KHz to 48KhHz.
 
Anything more complicated like you explained ~ 192000*147/640 = 44100, would mean extrapolation I guess. Then it comes down to quality of DAW, I guess.

 
Any modern DAW should be more than sufficient. An "echt" rational down-sampler acts with just an extra step compared to an integer one:
1) Repeat 0s between samples to get up to the common-multiple rate
2) Lowpass for your intended rate
3) Decimate samples to get to your intended rate
 
An integer down-sampler skips 1.

 

[spruce music]

Actually according to those that know more than I do about digital audio, and the math involved, Resampling to non-integer rates is superior by a small amount.  Isn't intuitive, but much of digital is like that.  Ever wonder why Benchmark DACs for a long time resampled everything to 110 khz?  That is why.  Non-integer conversion is better and rejects jitter that may be in the original bit stream.

 

[cjl]

Interestingly, even when converting to an integer rate, you don't just want to take every fourth sample. You can introduce aliasing if you do that, which can add very audible artifacts. What you actually want to do is lowpass it first, then resample, to ensure you have eliminated any frequencies above your new nyquist.

 

[Roseval]

  Ever wonder why Benchmark DACs for a long time resampled everything to 110 khz?  That is why.  Non-integer conversion is better and rejects jitter that may be in the original bit stream.

I’m afraid you are talking about something completely different.
 
The OP asks about converting a file to a file with a different sample rate and/or bit depth.
As this is from file to file, we do stay in the digital domain hence no jitter is added or removed.
 
You are referring to ASRC, asynchronous sample rate conversion, that what the DAC1 is doing. This is decoupling the clocking of the input stream from the clocking driving the DA.
Here indeed we talk about actual physical clocks.
ASRC is a powerful technique to get rid of input jitter in a DAC.

 

[Lex2]

   
Great. Informative. That's what I was hoping for. I have Audacity, but I am currently looking at DAWs. I am going to run both Cubase and Pro-Tools trail versions in the coming weeks. (Probably stating this week. I am wrangling the trial downloads around my current ISP cap.)
 
I used to own Cubase, and used both DAWs at university. That was while back though an we didn't have such high sampling rates then. Or not that I knew of. I think I recall 48K. (Anyway I am talking way back at the very end of the nineties.)
 
I am jumping back on the horse so to speak, and getting back on with my hobby. Thank you for helping. As long as I know it can be done, and easily, then I am OK.
 
I can't help thinking that doing direct multiples is better though. That way I guess the compression literally takes, e.g. the fourth sample in converting 192KHz to 48KhHz.
 
Anything more complicated like you explained ~ 192000*147/640 = 44100, would mean extrapolation I guess. Then it comes down to quality of DAW, I guess.

You may also want to check out Studio One Pro 3.3 (the trial version), developed by one of the ex-developers for Cubase. It's rock solid stable, although not as fully featured as Cubase 8 is now. But the project management options are a lot more user friendly than Cubase, and includes a mastering module. MIDI functionality has still a long way to go, however.

 

[RRod]

You can go here and see comparisons of SRCs on several metrics. You'll see some, um, interesting things occasionally.

 

[GreenBow]

Thank you for continued input. I read every post enthusiastically.
 
I am thinking the best thing to do might be to record in RedBook. Then I won't have to down-sample, if down-sampling causes jitter or any errors. However the methods should be safe because the pro0industry uses down-sampling. They absolutely record in high sampling frequencies than RedBook. 
 
@Lex2 I think I intend to stick to Cubase or Pro-Tools, because I know them from years ago. I also bought Cubase back in about 2001, but sold it later. I was having an issue with it, that I think was  entirely my fault.
 
I like also the score manuscript element of the Cubase Pro version. I have a music project I would use that for. While Pro-tools have Sibelius, I don't want to be spending money for another whole new add-on module.
 
That's why it's my plan to try both trial versions in the next few weeks. I might even be able to complete a project on the trial version of Cubase. Or find that the Lite free version would be sufficient to finfish a project or two. (Though the Lite version might not contain the ability to work score manuscript. I can check that online. However I have a project that doesn't need manuscript to complete.) 

 

[gregorio]

  That was while back though an we didn't have such high sampling rates then. Or not that I knew of. I think I recall 48K. (Anyway I am talking way back at the very end of the nineties.)
 
However the methods should be safe because the pro0industry uses down-sampling. They absolutely record in high sampling frequencies than RedBook. 

 
No, we did have higher sampling rates back then. It's worth knowing a bit about trends: At the end of the 90's we had higher sample rates but no one used them because no distributor wanted higher, there was no consumer demand and no point loosing all the additional computing resources required to process them. For roughly the first five years or so of the millennium, some ADCs/DACs operated slightly more transparently at 96kHz and a number of plug-in processors also worked better at higher than 48k, particularly non-linear, soft-synth and modelling plugins. Today most of those potential advantages are gone. While some plug-ins do still operate better at higher sample rates, this is handled by the plug-ins themselves up-sampling, processing and down sampling again, all internally, before passing the audio back to the DAW. Plug-in output quality has improved tremendously over the last decade and many who swore vehemently about their tinny, digitally harsh sound are now converts and their once treasured hardware units are gathering dust in a cupboard. Pro quality ADCs/DACs also no longer generally perform better at 96kHz and at no time in history has it ever been recommended to use 192kHz. Indeed, Pro Tools (DigiDesign/AVID), published a paper specifically advising against 192kHz, on audio quality grounds, rather than computing power requirements. Today "they absolutely" do not record at high sampling rates. Quite a few still record at 96kHz, either because they have an older ADC/DAC which performs better at that rate, or because they haven't kept up with developments in the last 5-10 years and don't realise there is no longer any logical reason for higher sample rates than 44.1/48k. The only pros who record at 192kHz are those employed to by record labels, to fulfil a demand for 192k some labels have helped create, otherwise, no one I know (or have heard of) would willing record at 192k.
 
Your down-sampling concerns are unfounded. They weren't so unfounded 15 or so years ago when it was generally recommended to avoid a non-simple multiples, but they are today. These days, unless you've got very old or incompetently programmed software, conversion from 192k to 48k or 44.1 should be equally transparent. In fact, it's extremely likely that any fairly recent commercial recordings you own have been up and down sampled multiple times, possibly even dozens of times, as part of the modern mixing process, as it's a standard built in procedure of many plug-ins, which btw is performed automatically in the background with the mix engineer not even aware it's occurring. My advice, if you're getting back into recording/mixing/producing is to stick with 44.1 or 48kHz, there's no benefit to the higher sample rates which isn't already accounted for. If you feel strongly about higher sample rates, 88.2 or 96k are fine, apart from the near halving of your resources of course, avoid 192k though, unless you have a very specific requirement!
 
G

 

[spruce music]

   
No, we did have higher sampling rates back then. It's worth knowing a bit about trends: At the end of the 90's we had higher sample rates but no one used them because no distributor wanted higher, there was no consumer demand and no point loosing all the additional computing resources required to process them. For roughly the first five years or so of the millennium, some ADCs/DACs operated slightly more transparently at 96kHz and a number of plug-in processors also worked better at higher than 48k, particularly non-linear, soft-synth and modelling plugins. Today most of those potential advantages are gone. While some plug-ins do still operate better at higher sample rates, this is handled by the plug-ins themselves up-sampling, processing and down sampling again, all internally, before passing the audio back to the DAW. Plug-in output quality has improved tremendously over the last decade and many who swore vehemently about their tinny, digitally harsh sound are now converts and their once treasured hardware units are gathering dust in a cupboard. Pro quality ADCs/DACs also no longer generally perform better at 96kHz and at no time in history has it ever been recommended to use 192kHz. Indeed, Pro Tools (DigiDesign/AVID), published a paper specifically advising against 192kHz, on audio quality grounds, rather than computing power requirements. Today "they absolutely" do not record at high sampling rates. Quite a few still record at 96kHz, either because they have an older ADC/DAC which performs better at that rate, or because they haven't kept up with developments in the last 5-10 years and don't realise there is no longer any logical reason for higher sample rates than 44.1/48k. The only pros who record at 192kHz are those employed to by record labels, to fulfil a demand for 192k some labels have helped create, otherwise, no one I know (or have heard of) would willing record at 192k.
 
Your down-sampling concerns are unfounded. They weren't so unfounded 15 or so years ago when it was generally recommended to avoid a non-simple multiples, but they are today. These days, unless you've got very old or incompetently programmed software, conversion from 192k to 48k or 44.1 should be equally transparent. In fact, it's extremely likely that any fairly recent commercial recordings you own have been up and down sampled multiple times, possibly even dozens of times, as part of the modern mixing process, as it's a standard built in procedure of many plug-ins, which btw is performed automatically in the background with the mix engineer not even aware it's occurring. My advice, if you're getting back into recording/mixing/producing is to stick with 44.1 or 48kHz, there's no benefit to the higher sample rates which isn't already accounted for. If you feel strongly about higher sample rates, 88.2 or 96k are fine, apart from the near halving of your resources of course, avoid 192k though, unless you have a very specific requirement!
 
G

Nice post, but you didn't address the need for 384 khz and 768 khz recording.  Plus I hear rumours from Japan that 1536 khz/32 bit sounds better still.