A compendium on SR and bit depth - Ableton's sound quality

[vitamin-e]

I am currently running tests with my new mix on the following options:

Export @ 41.1/16 with powr3
Export @ 41.1/16 with no dither
Export @ 192/32, re-import, dither with pow-r3 to 41.1/16
Export @ 192/32, re-import, no dither to 41.1/16
Export @ 96/32, re-import, dither with pow-r3 to 41.1/16
Export @ 96/32, re-import, no dither to 41.1/16
Export @ 88.2/32, re-import, dither with pow-r3 to 41.1/16
Export @ 88.2/32, re-import, no dither to 41.1/16

Once complete, I will convert all of the 41.1/16 clips to 320kbps mp3, upload them to my site, and have you guys tell me what sounds best. After much deliberation, I've decided not to use FLAC or uncompressed WAV, as it would be too big, and most Ableton musicians are releasing their tracks via 128kbps download anyway...

I also am going to try playing 96/24 analog into a 41.1/16 ADC and recording the re-digitized result - I have to wait til I get back into the studio for this though

I am also going to run phase cancellation tests to see the difference between the files (the super hq vs the downsampled/dithered final product.) Obviously a re-sampled, dithered, file will NEVER zero in a phase cancellation test - the test is intended to indicate how much it changed - presumably, information lost by the conversion is what we will see on the analyzer - that and dithering noise.
EDIT: What was I thinking? This is impossible because Ableton will automatically convert the SR (at the point in the signal chain where the sound leaves the file and would normally go to the fx chain if present) to the one specified in preferences - so I'd be comparing Ableton's realtime sr conversion to Ableton's sr conversion at the time of export. That's for another day.
I guess this test will be purely a listening experiment...

What I can do is run a phase cancellation test with 41.1/16 powr3 dither vs a 41.1/32 export, reimported and dithered to 41.1/16. This will confirm or deny my hypothesis that Ableton is forcing my plugins into an inferior bitrate when exporting - rather than using the highest possible, and dithering at the output.


Also, for anyone wondering - I'm not bashing Ableton's sound quality - I fucking love this program more than any other DAW(or studio equipment - analog or digital) I've ever used. I don't feel like Ableton's sound quality is an issue, though I'm sure that there is a "best" way to preserve every minuscule detail your final product for the end listener, and I'm sure it's a fine balance of several things. I'm just an ultra-perfectionist with time on his hands.

[pepezabala]

I am currently running tests with my new mix on the following options:

Export @ 41.1/16 with powr3
Export @ 41.1/16 with no dither
Export @ 192/32, re-import, dither with pow-r3 to 41.1/16
Export @ 192/32, re-import, no dither to 41.1/16
Export @ 96/32, re-import, dither with pow-r3 to 41.1/16
Export @ 96/32, re-import, no dither to 41.1/16
Export @ 88.2/32, re-import, dither with pow-r3 to 41.1/16
Export @ 88.2/32, re-import, no dither to 41.1/16

Once complete, I will convert all of the 41.1/16 clips to 320kbps mp3, upload them to my site, and have you guys tell me what sounds best. After much deliberation, I've decided not to use FLAC or uncompressed WAV, as it would be too big, and most Ableton musicians are releasing their tracks via 128kbps download anyway...

This is the same as saying: "I scan an image at various resolutions, and then print it at various resolutions.

I will then compare the results looking at them with sunglasses. Most people wear sunglasses anyways."

I mean you also could say "I will then compare the results after a couple of drinks and some pills in a crowded bar on a shitty soundsystem, through an overdriven dirt cheap dj-mixer" - this is probably the average scenario for electronic music, isn't it.

[Leon Tricker]

...with time on his hands.

You are very lucky, in that case.

If I had that luxury I'd spend it in a completely different way.

[vitamin-e]

I've finished exporting all of the files, and uploaded them. Gotta get to sleep, and will probably run phase cancellation tests tomorrow - though it would appear that there is some randomness to the track I'm working with since I'm using all synths and samplers, and no dumped tracks. I'll have to figure another way to go about this - probably with sweeps or impulses instead of a song. This is intended to be more of a listening test. I was thinking of flattening the tracks, but that would do no good for this test, and I'm not sure how Ableton treats flattened tracks at the time of export.

Download the .zip at http://ericseastrand.com/export_tests/export_tests.zip

Listen to individual MP3s at http://ericseastrand.com/export_tests/

[melocoton]

Before reading, please understand:

Almost everything you wrote about digital audio is totally false. Before going any further you should try to understand that digital audio accurately reproduces sound up to a frequency of 1/2 of the sample rate. Increasing the sample rate captures higher frequencies and increasing the bit depth captures a greater dynamic range, but neither of those factors increase detail in the range of sound that was already captured. I think you should understand how and why that works before continuing on with all of these tests.

[crumhorn]

I've finished exporting all of the files, and uploaded them. Gotta get to sleep, and will probably run phase cancellation tests tomorrow - though it would appear that there is some randomness to the track I'm working with since I'm using all synths and samplers, and no dumped tracks. I'll have to figure another way to go about this - probably with sweeps or impulses instead of a song. This is intended to be more of a listening test. I was thinking of flattening the tracks, but that would do no good for this test, and I'm not sure how Ableton treats flattened tracks at the time of export.

Download the .zip at http://ericseastrand.com/export_tests/export_tests.zip

Listen to individual MP3s at http://ericseastrand.com/export_tests/

Your file names give the game away. For this listening test to have any value you should give the files random names.

[Anubis]

I am currently running tests with my new mix on the following options:

Export @ 41.1/16 with powr3
Export @ 41.1/16 with no dither
Export @ 192/32, re-import, dither with pow-r3 to 41.1/16
Export @ 192/32, re-import, no dither to 41.1/16
Export @ 96/32, re-import, dither with pow-r3 to 41.1/16
Export @ 96/32, re-import, no dither to 41.1/16
Export @ 88.2/32, re-import, dither with pow-r3 to 41.1/16
Export @ 88.2/32, re-import, no dither to 41.1/16

Once complete, I will convert all of the 41.1/16 clips to 320kbps mp3, upload them to my site, and have you guys tell me what sounds best. After much deliberation, I've decided not to use FLAC or uncompressed WAV, as it would be too big, and most Ableton musicians are releasing their tracks via 128kbps download anyway...

This is the same as saying: "I scan an image at various resolutions, and then print it at various resolutions.

I will then compare the results looking at them with sunglasses. Most people wear sunglasses anyways."

I mean you also could say "I will then compare the results after a couple of drinks and some pills in a crowded bar on a shitty soundsystem, through an overdriven dirt cheap dj-mixer" - this is probably the average scenario for electronic music, isn't it.

That's why I always do my final mixes on headphones, because kids are gonna be listening to my shit on those crappy iPod ear plugs.
(OK, I really don't think that any "kids" would listen to my "shit" anyways, it's too depressing)

[silveriofunk]

because people want to find excuses for other people not liking their music. "oh must be the sound engine, thats why i'm not at no.1 " or " I almost made it as a pop star, but i was using ableton and the sound wasnt good enough"

Oh, I see.

My music sucks because I lack inspiration and musical talent. It has nothing to do with Ableton's sound engine.

when your music sucks it sucks....

when it rocks it rocks

simple huh?

[supamonsta]

most Ableton musicians are releasing their tracks via 128kbps download anyway...

we're talking about audio quality right here, in a kind of gearslutz manner (I like it), and then you seamlessly drop this (quote) !!!!

I'd like to know WHO releases its music as 128K mp3s!!!

I'm sure NO ONE interested in this topic does that.

Anyway I find this thread interesting, I'm downloading your zip to give it an ear.

Cheers

[Silence]

come on, the dude makes en effort not to bash Live and ppl are being mean.
i for one would love a "best practice" thread. Music might be a very individual thing but knowing the way a tool works wouldn't hurt now would it?

[leedsquietman]

I sincerely doubt that anyone exports their music at 128 Kbps anymore, For the last 5 years it's been known that 128 Kbps is not as 'near CD quality' as once described, and more and more sites allow 320 Kbps for download, upload, purchase and also .wav or lossless FLAC etc.

However, sites like MySpace, Soundcloud, Bandcamp, Soundclick and most online music distributors do downsample the previews and streaming down to 128 Kbps (and possibly lower). Soundcloud, Bandcamp and others do allow the downloads to be at the higher bitrate/.wav/.flac - so when listening to streaming audio, previews etc.

Even on Ipods and MP3 players, most people are getting at least 192 Kbps mp3 quality (or 160 Kbps aac which is equivalent quality). This is a small but noticeable difference if your music is layered and has a lot of frequency and dynamic range.

[dinaiz]

come on, the dude makes en effort not to bash Live and ppl are being mean.
i for one would love a "best practice" thread. Music might be a very individual thing but knowing the way a tool works wouldn't hurt now would it?

I second that

[supamonsta]

it is in fact a really interesting question, how to optimise live's sound with correct workflow.

I don't see any really mean people in this thread, apart those who misread the OP at the early beginning (as I actually did the first time) and that were bashing this thread as another "DAW sound engines comparison" topic.

Music might be a very individual thing but knowing the way a tool works wouldn't hurt now would it?

agree. it's a must-do. And I'm interested by this question. Already learned here and there that real-time rendering (recording the master in a stereo track, no master rendering) was much better as "master rendering" because of the fact that Live's realtime-processing engine did something special to the sound, that did not happen when rendering. So, to get a final wav file that was 100% accurate with the sound listened to while mixing, it would be better to record than to render.

tips like this may be gold.

there are also many ways of improving the sound by the way we mixdown our tracks, and lots of useful advices and tutorials can be found in this forum.

A particular Sticky or forum section would be nice, yes.

cheers

[locojohn]

Bon jour!

...Already learned here and there that real-time rendering (recording the master in a stereo track, no master rendering) was much better as "master rendering" because of the fact that Live's realtime-processing engine did something special to the sound, that did not happen when rendering. So, to get a final wav file that was 100% accurate with the sound listened to while mixing, it would be better to record than to render.

Pardon me, are you saying that rendered Master WAV at project sample rate and bit depth, with Normalize and Dither OFF sounds different than when you hear it playing in Live project? If so, this is very new to me, as I never had this observation before. Please kindly clarify.

Andrejs

[davepermen]

Bon jour!

...Already learned here and there that real-time rendering (recording the master in a stereo track, no master rendering) was much better as "master rendering" because of the fact that Live's realtime-processing engine did something special to the sound, that did not happen when rendering. So, to get a final wav file that was 100% accurate with the sound listened to while mixing, it would be better to record than to render.

Pardon me, are you saying that rendered Master WAV at project sample rate and bit depth, with Normalize and Dither OFF sounds different than when you hear it playing in Live project? If so, this is very new to me, as I never had this observation before. Please kindly clarify.

Andrejs

haven't looked at it precisely, but i know that export does it with a single core, while realtime does use multiple cores. there are differences between those two modes (documented in the .. documentation).

but i don't know if this is actually hearable.