Buggy REs with more than 44.1khz

[Loque]

Is there a list or any information which RE supports higher resolutions than 44.1 khz?

I noticed several RE producing different results or just crash. Eg Carve crashes sometimes and Echo Bode produces different results in 98khz with 24 bit. Looks like some developers did not checked that and in the manuals i could not find any information about support of higher resolutions?

[JiggeryPokery]

All RE's support resolutions higher than 44.1, except Uhbik-A.

[Loque]

Well, i just checked out the song part i was working on yesterday and a quick investigation shows, that at least Neutron Particle Emitter goes out of sync on 96khz/24bit. See attached songfile for an example (remove .txt at the end).

But, i can remember that after rendering a song a while ago a complete track was missing if rendered >44.1khz. I did not spent the time for investigation, but finally i was wondering that i did not received any feedback from Reason. After several tries Carve crashed 2 times. And than i stoped a deeper investigation, because the rendering in full resolution took a lot (!!!) time and i did not wanted to disable each plugin and try again...

[JiggeryPokery]

Well, i just checked out the song part i was working on yesterday and a quick investigation shows, that at least Neutron Particle Emitter goes out of sync on 96khz/24bit. See attached songfile for an example (remove .txt at the end).

But, i can remember that after rendering a song a while ago a complete track was missing if rendered >44.1khz. I did not spent the time for investigation, but finally i was wondering that i did not received any feedback from Reason. After several tries Carve crashed 2 times. And than i stoped a deeper investigation, because the rendering in full resolution took a lot (!!!) time and i did not wanted to disable each plugin and try again...

Yes, an RE can still have a bug, but technically they all support all resolutions above 44.1.

Higher resolutions will often sound slightly different, for example, there is less aliasing. But that equally applies to Thor, and so this is not RE specific.

[Loque]

Ok. Than all RE supports all resolutions. And RE still can have bugs in different resolutions that i might notice atonce or later. And there is no guarantee that all REs are testet in all resolutions. This reminds on bad VST days in the past...

Than i only can hope most developers test their RE in different resolutions. Fingers crossed!

[electrochoc (PRX-A)]

Well, i just checked out the song part i was working on yesterday and a quick investigation shows, that at least Neutron Particle Emitter goes out of sync on 96khz/24bit. See attached songfile for an example (remove .txt at the end).

But, i can remember that after rendering a song a while ago a complete track was missing if rendered >44.1khz. I did not spent the time for investigation, but finally i was wondering that i did not received any feedback from Reason. After several tries Carve crashed 2 times. And than i stoped a deeper investigation, because the rendering in full resolution took a lot (!!!) time and i did not wanted to disable each plugin and try again...

Yes, an RE can still have a bug, but technically they all support all resolutions above 44.1.

Higher resolutions will often sound slightly different, for example, there is less aliasing. But that equally applies to Thor, and so this is not RE specific.

One day, I use Malström LFOs as oscillators (because why not? ), and noticed that the sounds yielded by that setting were very, very different at different resolutions! But I guessed at the time it was normal: a higher resolution means a higher precision, and more precision, in extreme settings, may add a lot of information for the computer to manage when it turn it into sounds, to the point that it can change the sound pretty much. I suspect that might be what the OP noticed with Echobode...

[DLDTech]

It's quite tricky to get DSP to work exactly the same at different sample rates, in reality you're always going to get some differences, so the best bet is to export at whatever sample rate you were doing your final mixing at.

One other gotcha is that most audio DSP code deals with chunks of n samples at a time, and it's common do to only compute non-audio frequency values once every chunk (LFO, Envelopes etc). At 44kHz and 48kHz, these chunks come at very similar rates, so you're unlikely to notice any difference, but at 96 or 192kHz, then you may get radical differences especially if the coder hasn't taken sample rate into consideration with non-audio values.

Personally I do most of my testing at 44 and 48kHz, and then do final testing at 96 just to make sure nothing is obviously wrong.

[dioxide]

It's quite tricky to get DSP to work exactly the same at different sample rates, in reality you're always going to get some differences, so the best bet is to export at whatever sample rate you were doing your final mixing at.

One other gotcha is that most audio DSP code deals with chunks of n samples at a time, and it's common do to only compute non-audio frequency values once every chunk (LFO, Envelopes etc). At 44kHz and 48kHz, these chunks come at very similar rates, so you're unlikely to notice any difference, but at 96 or 192kHz, then you may get radical differences especially if the coder hasn't taken sample rate into consideration with non-audio values.

Personally I do most of my testing at 44 and 48kHz, and then do final testing at 96 just to make sure nothing is obviously wrong.

Wouldn't some kind of oversampling help with this problem?

[DLDTech]

Wouldn't some kind of oversampling help with this problem?

Upsampling to 192 then downsampling back to native frequency would solve all these issues, but you'd waste a huge amount of DSP time doing this.

In reality, you only normally oversample stuff that really needs it as it's a killer on CPU, and often oversampling really means 'run this bit of code a few times until the output values settle'.

And of course, the final audio is likely to be released at 44k anyway, then upsampled to 192k or similar by the audio DAC to avoid aliasing.

[joeyluck]

Just curious, in what applications do folks want to use higher rates than 44.1 kHz?

[avasopht]

Just curious, in what applications do folks want to use higher rates than 44.1 kHz?

FM and distortion for a start, which will introduce harmonics above nyquist, thus introducing aliased frequencies.

[DLDTech]

Well those are the sort of applications where oversampling should occur internally, and then downsample to 44k after filtering out anything above 20kHz. There's no need to run the entire rack at 96k for this.

[avasopht]

Well, unless you want that sound

[Loque]

Just curious, in what applications do folks want to use higher rates than 44.1 kHz?

Thank god i normaly finish my songs before DSP is too high, but there is not enough DSP for mastering in most cases. Thus, I create a new mastering-song (or bounce) only for mastering processing. Therefore i wanted to have the highest possible resolution - that was the spot i found the problems.

Due to the problems with the REs i came down to another solution. Now i render the song (or bounce) in 44khz, master it, than copy the mastering stuff to the original song and render it. This way i dont have quality lost or dithering problems. But realy...this is working around bugs...

Another point when i have have the problems with the sample rate is, when i render some samples to use it later in other productions or just because they use too many effects. Than i wanted to have a high resolution for the song i use the sample with, but the problems with the RE forces me to use only 44khz.

Hope this makes my uses-cases a bit clearer.

[JiggeryPokery]

Just curious, in what applications do folks want to use higher rates than 44.1 kHz?

FM and distortion for a start, which will introduce harmonics above nyquist, thus introducing aliased frequencies.

Yes, higher sample rates will cure that, but you don't really need to run Reason beyond 48kHz. Users can simply select 96kHz in the render dialog box when running Export Song, and Reason will use that rate internally rather than the external (i.e., ASIO) selected system rate. With export live playback isn't necessary as rendering is not done in real-time.

@ loque - If mastering later you should always export at 96/24, there's no reason at all not to.

[gak]

Just curious, in what applications do folks want to use higher rates than 44.1 kHz?

FM and distortion for a start, which will introduce harmonics above nyquist, thus introducing aliased frequencies.

Maybe, but just as an example, the bug I've reported (and been confirmed though no fix ) is present regardless of sample rate. (absolutely no improvement)

Maybe that's not what you are saying, but in practice I've found that as long as your card has decent converters, it's not really noticeable.

I just got tired of the game "will the cpu be ok?" (organ) "will the plugin work" (ORGAN) "will sally be staying with Phil overnight (glorious old school organ playing the theme to the next commercial )

[Benedict]

Yes many DSP devices (RE or VST alike) will deliver different results at different rates. Up to you if that is better or worse based on your needs. I made my last record at 24/96 and had to render at that rate and convert back to 16/44 later (exporting at 16/44 sounded different from what I heard so take care with monitoring at one rate and exporting at another. I only worked at 96k because I wanted to use some fast modulations and FMs which were smoother at the higher rate. This time I am back to 16/44.