What does ableton do when I playback higher sample rates at:

[physicalpatrick]

What happens in ableton when I play a clip that was recorded at 96khz back when the project is set at 48khz?
Does it 'resample' audio? Or quantize the audio? Are the extra samples ditched or are they influencing the ones that are being heard? Is there a difference between playback live and rendering out?

What about if a 96khz sample is playing back in a sampler under same set of circumstances?

I don't want to start an argument about whether or not I need to do this I just want to know the answer to the question.

Ableton software engineers are welcome to comment.

Thanks in advance!

[jestermgee]

While you sit and wait for replies, Google suggests:

viewtopic.php?f=1&t=142333

As a start (sorry, I only have 0.56 seconds to chuck in an answer).

[michaelandrews]

My understanding of this is that sample rates usually aren't "converted" and played back at the other sample rate on-the-fly, they're pitch-shifted up or down. A file with a sample rate of 44.1kHz has LESS samples than one of 96kHz. As such, squeezing 2.5x (or whatever it ends up being) the samples into the same playback time period means you have to speed the sample up, which shifts the pitch up (for a real-world example of how this can go horribly wrong, see http://popwatch.ew.com/2007/10/29/van-halens-jump/).

To get around this, I'm fairly certain that Ableton converts files from their native sample rate to the DAW sample rate, as it does when working with compressed (MP3, etc) audio and makes a WAV copy of it.

From the Live8 manual (http://downloads.ableton.com/manuals/81 ... ual_en.pdf):

31.3.2 Sample rate conversion/transposition

Sample rate conversion (during both real-time playback and rendering) is a non-neutral
operation. Playback of audio les at a sample rate that is different from the rate set in
Live's Preferences window will cause signal degradation. Transposition is also a form of
sample-rate conversion, and thus also results in non-neutral behavior.

To minimize potential negative results, it is recommended to do sample rate conversion as
an ofine process in another application. Once the samples have been converted to theHAPTER 31. AUDIO FACT SHEET 514
sample rate that you plan to use in Live, the les can be imported without any loss of quality.
Rendering audio from Live with a sampling rate other than the one that was used while
working on the project is also a non-neutral operation, and may result in a loss of sound
quality. It is recommended to always render using the original sampling rate, and then
convert the rendered le using a dedicated mastering application that is optimized for
these kinds of CPU-intensive, ofine tasks.

While we recommend that you use a high-quality ofine tool for sample rate conversion,
we recognize that one of Live's core features is its ability to pitch-shift and warp audio in
real time. For this situation, it is necessary to make a trade-off between CPU performance
and precision. We recommend the use of the Hi-Q button for any clips which undergo
transposition in a given Set. The algorithm behind the Hi-Q switch was rewritten for Live 7,
and now results in considerably lower distortion than in previous versions.

Also worthwhile from the manual is this section:

31.4 Tips for Achieving Optimal Sound Quality in Live

For users looking to achieve optimal audio quality in Live, we have provided a list of
recommended practices and program settings.

? Decide which sample rate to use for a project prior to beginning work, rather than
changing the sample rate while working on the project.

? Record audio into Live using high-quality hardware components (audio interface, cables, etc.) and at the highest sample rate and bit depth your interface and computer
will support.

? Avoid using samples that are at different sample rates within the same project. If you
want to work with such les, we recommend that you rst convert them to the sample
rate set for your audio interface in an ofine application that is optimized for this task.

? For all audio clips, disable both the Warp and Fade options in the Clip View.

? Do not adjust the Transpose and Detune controls for any clips.

? Always render at 32-bit and at the sample rate set for your audio interface. If you
need audio les at a different sample rate and/or bit depth, we recommend that youCHAPTER 31. AUDIO FACT SHEET 517
convert your rendered les in an ofine application that is optimized for these tasks,
rather than in Live.

Please note that these practices, while ensuring optimal audio quality, disable some of Live's
functionality in particular, stretching and synchronization.

Hope this helps!

[physicalpatrick]

Thanks for all the insight...Very Helpful!

What if i'm using Sampler?

Here's my workflow....I typically work in 48k so I switch to 96k just to record some fat ass samples from the Arturia Minimoog synth pumped through some distortion. Then I drop these clips onto the sampler and switch back to 48k and start playing.

So how are they playing back in this scenario???

[jestermgee]

Playback of audio at a sample rate that is different from the rate set in
Live's Preferences window will cause signal degradation.

Your samples will be played back at the lower samplerate of 48k.Wether you can hear the difference will be up to you but it will depend on the information in the material and Abletons algorithm for resampling.

This is something I believe gets people when they work within Ableton at high samplerates (above 44.1k) then just render their project to 44.1k. Without the proper care and dither settings (of which Abletons in-built dithering is ok but not ground breaking) the final result will come out sounding "flat" or empty (void of space).

Do some research on what Dithering is and why its used. The concepts apply to all digital audio and help explain what is happening (but be warned it's kind of a tricky area to understand.

[Tarekith]

Keep in mind that because Live can run multiple samplerates in the same project, and is meant to do so live, that the sample rate conversion it uses isn't the greatest. It's not terrible or anything, but if you're going to go to the trouble of recording at 96k, you probably want to downsample with something else before bringing it into Live. I know myself and a lot of people I know use Sample Manager from Audiofile for this, works great.

I generally recommend people just work at 44.1kHz the way through if they are working solely in Live.

[73*]

Do some research on what Dithering is and why its used. The concepts apply to all digital audio and help explain what is happening (but be warned it's kind of a tricky area to understand.

I don't mean to sound rude, but I would highly advise you follow your own advice. Dithering has absolutely nothing to do with sample rate conversion, it is a processed applied to bit reduction, and not tied to sample rate conversions.

If you are doing any processing that involves sample rate conversions and bit reduction, first do the SRC at the sources original bit depth (ie: highest available), then apply dither during bit reduction.

[jlgrimes]

What happens in ableton when I play a clip that was recorded at 96khz back when the project is set at 48khz?
Does it 'resample' audio? Or quantize the audio? Are the extra samples ditched or are they influencing the ones that are being heard? Is there a difference between playback live and rendering out?

What about if a 96khz sample is playing back in a sampler under same set of circumstances?

I don't want to start an argument about whether or not I need to do this I just want to know the answer to the question.

Ableton software engineers are welcome to comment.

Thanks in advance!

I don't think u will have problem going from 96 to 48. That is a clean conversion where all Ableton has to do is skip every sample. The problem is going to a sample rate that isn't an integer multiple of the other.

So 96 to 44.1 would be an issue.

88.2 to 44.1 wouldn't.

My advice is if u are primarily working for CD or mp3 releases use 44.1. If u desire a higher rate use 88.2.

48 and 96 kHz is more for film.

[Tone Deft]

my advice is to take some time and try all these things for yourself.

this forum has gotten a lot better at not falling for old mistaken cliches but I highly encourage anyone with questions on this stuff to do some experiments on this stuff. approach it with the right mindset and it's fun.

[physicalpatrick]

I do alot of graphics work and a dumbed down version of dithering is like adding film grain to computer generated graphics... it most noticably reduces banding on subtle gradients...and can help 'warm up' the image.

if you play a 96khz sample across the keyboard obviously every key has a different sample rate going in but how does it come out the other end as 48k (my project setting)? are the extra samples tossed in the bin? or factored into the audible result?

pat

[Tarekith]

I don't think u will have problem going from 96 to 48. That is a clean conversion where all Ableton has to do is skip every sample. The problem is going to a sample rate that isn't an integer multiple of the other.

So 96 to 44.1 would be an issue.

88.2 to 44.1 wouldn't.

.

That is not at all how sample rate conversion works these days.

[timday]

I don't think u will have problem going from 96 to 48. That is a clean conversion where all Ableton has to do is skip every sample. The problem is going to a sample rate that isn't an integer multiple of the other.

So 96 to 44.1 would be an issue.

88.2 to 44.1 wouldn't.

.

That is not at all how sample rate conversion works these days.

Doesn't it? I thought that when converting non integer sample rates the values of the new samples were calculated using the sinc function of the original samples, but those calculated samples were subject to quantisation error? Have I misunderstood or has it changed?

Not trying to be arsey, it's just that what I thought I knew may be wrong.

[Tarekith]

My understanding is that since any decent SRC up samples (over samples?) there's no such thing as one SR being "easier" or better when it comes to the conversion.

[jlgrimes]

My understanding is that since any decent SRC up samples (over samples?) there's no such thing as one SR being "easier" or better when it comes to the conversion.

I guess we don't know what method Live is using. I know that back in the day a good sample rate conversion cost a lot of CPU cycles, similar to time stretching. But now most time stretching can be done in real time because of faster CPUs.

I'm suspecting though Ableton might have possibly cut corners here because Live is a DAW that is optimized for real time performance. I also think that them stating to pick a sample rate and stick with it in your projects also imply that they might not be using the best conversion. They also state to use a dedicated program to do sample conversions if mastering is the intent.

I would only feel safe using a sample rate that is double or quadruple the final rate because the math would be a lot simpler.

[73*]

This site will be helpful to anyone wanting to see proper comparisons between the different SRC options available...

http://src.infinitewave.ca/

It is worth everyone noting that Voxengo R8brain free out performs many of the SRCs in DAWs and other paid-for options, of course it's not realtime, but trying to do high quality realtime SRCs is arguably a huge waste of CPU resources. Hence Ableton's suggestion of sticking to a single sample rate within projects.