From the user guide - "Opinion is split on whether working at higher rates is worthwhile, but if you like to work at higher sample rates then it is a good idea to increase this value in fixed multiples of your desired final sample rate (e.g., 88200 Hz for CD audio which will end up at 44100 Hz)"
So 44.1 is better than 48?
Does this take into account the bit depth? Say I record at 48k-24 bit and have to render to 44.1K-16 bit. Wouldn't a render from 44.1k-24 bit to 44.1khz-16bit have the same issues?
Who wrote this into the guide anyway and what is it based on, general theories, or the actual way the Tracktion code implements the render?
Scott
lower sample rate is better?
-
- KVRist
- 152 posts since 20 Dec, 2003
-
- KVRAF
- 3364 posts since 16 Feb, 2004 from atop a katamari
bit depth isn't really related; you should always record at as high a bit depth as you can be bothered with; the extra precision can always be thrown away without much issue. that's related to dynamic range, and you can get far greater range of 'levels' if you have a bigger slot to store the number in.
Kick, punch, it's all in the mind.
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
If you are burning to CD, yes.DWS wrote: So 44.1 is better than 48?
Bit depth is unrelated.Does this take into account the bit depth?
No. I can explain the math if you like, but basically you can think of 24bit as being a direct multiple of 16bit, in the same way the 88.2 is a direct multiple of 44.1.Say I record at 48k-24 bit and have to render to 44.1K-16 bit. Wouldn't a render from 44.1k-24 bit to 44.1khz-16bit have the same issues?
The two things are indentical. Irrespective of host, the fundamentals of resampling stay the same.what is it based on, general theories, or the actual way the Tracktion code implements the render?
Someone shot the food. Remember: don't shoot food!
-
- KVRAF
- 1615 posts since 28 Mar, 2005
Hi Valley - there are two sides to the issue: On reason to use 48 is that it can push the necessary filters in the input device (and their artifacts) up out of the audible range. This has to be traded off against the loss from the ultimate conversion. I would offer that as a rule of thumb if your inbound converters are not great that 48 might sound better - I use 44.1 as my converters sound good there. certainly there is practically no reason to use 96 rather than 88.2. Finally there is the question of mastering in the analog domain: if you plan to do this, there is no penalty to using 48 rather than 44 - again it boils down to a comparison of the quality of the filters in your input device.
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
If the filters are that poor, you'll hear them at 48k too. The filters will slope long before the cut off, so you are still going to hear them working, maybe to a slightly lesser degree, but not enough to warrant a resample from 48 down to 44.1 k.
Someone shot the food. Remember: don't shoot food!
-
- KVRAF
- 1615 posts since 28 Mar, 2005
Your more or less right Valley I think the extra fifth of an octave could make a difference. It was more of an issue with early crap converters - oversampling renders it moot but there it is anyhow. re 88 who knows you might want a high rate master for the day when CDs are superceded!
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
I tend to think where hardware is concerned, all bets are off.
Your point abouyt losing the filters is good, but the ideal fix is to record at 88K so the filters are doing their work *way* outside of human hearing - then downsampling. However, then you need to balance the noisfloor and jitter performance of the converters at 88 against the performance at 44. I think many people assume that they'll be broadly the same, but all the tests I've seen suhghest that noise, stereo cross talk, frequency response, and jitter performance can vary sometimes dramatically between different sampling rates.
In general, I think that unless you have done heavy testing of your equipment, the *safe* option is to stick to multiples of your desired target sample resolution/rate. That's why I suggested people do that. If you know enough to break those laws, you know enough not to need to read that section anyway.
Your point abouyt losing the filters is good, but the ideal fix is to record at 88K so the filters are doing their work *way* outside of human hearing - then downsampling. However, then you need to balance the noisfloor and jitter performance of the converters at 88 against the performance at 44. I think many people assume that they'll be broadly the same, but all the tests I've seen suhghest that noise, stereo cross talk, frequency response, and jitter performance can vary sometimes dramatically between different sampling rates.
In general, I think that unless you have done heavy testing of your equipment, the *safe* option is to stick to multiples of your desired target sample resolution/rate. That's why I suggested people do that. If you know enough to break those laws, you know enough not to need to read that section anyway.
Someone shot the food. Remember: don't shoot food!
-
- KVRAF
- 1615 posts since 28 Mar, 2005
controversies aside, I actually record at 44.1 myself 
-
- KVRAF
- 4644 posts since 28 Nov, 2002 from Chicago
me too.
Someone shot the food. Remember: don't shoot food!
-
- KVRAF
- 10815 posts since 26 Nov, 2004 from UK
-
- KVRist
- 93 posts since 30 Jan, 2003
I'm a believer in 48k. My feeling is that since I use software samplers, and since they transpose up a little from their recorded pitch, that 48k gives me a little transposition range before aliasing.
-
- KVRist
- Topic Starter
- 152 posts since 20 Dec, 2003
What if 'I' am not burning to CD but someone else is (maybe a mastering engineer) and I don't know what processing software they are using to hopefully make my tracks sound better before they are 'burned' at 44.1. What you still suggest 44.1?If you are burning to CD, yes.
Why is 48 so prevelant anyway and why was it introduced when it's only ~4 more khz and the majority of music ends up on CD?
-s
-
- KVRAF
- 1615 posts since 28 Mar, 2005
Speak to your mastering engineer about their preference. If you are going to pass through the analog domain and then resample it then it is a matter for your engineers and his/her preferences re: his/her converters.
I mastered a CD a few years ago w/ Joe Gastwert and he preferred the sound of his best converters @ 44.1 to 44.8. At that time I did not have a high sample rate option so don't know how he feels about 96!
If the tracks are being mastered in the digital domain then 44.1 will be better than 48 - 88.2 *might* have some advantage (maybe ringing of fir filters would be rendered more accurately in the time domain and that extra computation *might* give you some tiny extra degree of accuracy when you downsample - maybe someone out there knows better...)
48 was invented in the early days of the technology to get the nyquist frequency and the attendant filters tuned to it out of the audible spectrum ie 24khz rather than 22.1 - hearing petering out around 20.... Oversampling rendered that advantage moot but you still have a fifth octave of range in the freq domain and thats worth something...
I mastered a CD a few years ago w/ Joe Gastwert and he preferred the sound of his best converters @ 44.1 to 44.8. At that time I did not have a high sample rate option so don't know how he feels about 96!
If the tracks are being mastered in the digital domain then 44.1 will be better than 48 - 88.2 *might* have some advantage (maybe ringing of fir filters would be rendered more accurately in the time domain and that extra computation *might* give you some tiny extra degree of accuracy when you downsample - maybe someone out there knows better...)
48 was invented in the early days of the technology to get the nyquist frequency and the attendant filters tuned to it out of the audible spectrum ie 24khz rather than 22.1 - hearing petering out around 20.... Oversampling rendered that advantage moot but you still have a fifth octave of range in the freq domain and thats worth something...
-
- KVRAF
- 16154 posts since 2 Dec, 2003 from Nashville, TN
Regardless of sample rate, when you go to a mastering engineer, they usually will love you if you give them the higher bit depth. It gives them more headroom to apply their "magic." With sample rate, one thing that is an advantage of using multiples is because it's less math that has to be done. It's basically taking out every other sample, as opposed to every "____" so many samples. This CAN cause more smooth transitions and requires easier math for the CPU to perform.
Koolkeys
Koolkeys
My host is better than your host
-
- KVRAF
- 16154 posts since 2 Dec, 2003 from Nashville, TN
Also, 48 was made origianally for DAT machines to compete with and give better sound than, a CD. And they didn't want it to be compatible as easy. Kind of shady, I think!
Koolkeys
Koolkeys
My host is better than your host