I think it sounds really bad, honestly. I was looking forward to not worrying about my loops’ BPMs like I could with the Octatrack, and use them anywhere. But the Stretch Machine doesn’t preserve the pitch at all really, all my loops sound out of tune when they’re in a different BPM. I’ve been listening to the loops playing in the Stretch Machine while changing the pattern BPM to listen to how it changes, and I can hear a tone generated from the stretch algorithm repeating grains of audio at whatever rate is doing it.
I’m actually interested in the DT2 timestretch algorithm because I’m hoping it is similar to the old-school MPC timestretch that sounds lofi (and super cool for vocals). Can anyone compare the two?
I want to know if ‘bad’ means ‘the granular artifacts are not drugged out enough for my perverted musical taste’ or ‘the algorithm does not provide clean & casual DAW-like time-stretching’?
Could someone post some audio clips of the same samples but on each sampler please? I’d be interested in hearing exactly what the differences are.
Actually I’d be pretty interested in a big “Timestretch algo comparison thread” where everyone can use the same samples on their devices and post the results, so we can really get an idea of the different approaches used and how it’s changed over the years.
You think this feature that people have been asking for since the original DT came out, which the OT has been able to do since the start, is meant to be used as an effect? I simply don’t believe that.
Perverted? Yeah my music taste is so perverted because I expected a better time stretching algorithm from a brand new device in the year 2024 from a company that already provided a serviceable algorithm on one of their devices 15 years ago. I did not expect an algorithm that I could find in Logic, but frankly that is not much to ask for considering the fact that it is not difficult to implement with highly up to date hardware that the DT2 is platformed on, and again, the year is 2024. But I seem to not be the only one with such perversions, imagine that.
What @basic_helix is getting at is that they and many others like myself don’t use timestretch to make premade loops work at whatever tempo/pitch you want where the goal is transparency and as few artifacts as possible. We generally use them as a sound design effect where the gross artifacts are a plus, hence the “perversion” whereas you want clean normal vanilla DAW like timestretch, which is valid as well. IMO even the Octatrack timestretch is pretty wonky and cool for artifact farming, though it may work for your use case
Yeah I know, you can use any time stretch algo to get artifacts for sound design if you stretch it enough. The werp machine was meant as a rough interpretation of time stretch and is amazing for sound design. An actual time stretching machine should be able to preserve the transient and pitch integrity of a sample within ±5 bpm, especially on such a formidable device as the DT2, and especially seeing as you can record quantized loops in the DT2. I want to be able to use the quantized loops that I can record into the DT2 in different contexts without having to decide the BPM of the pattern before I start working on it. I just don’t believe that Elektron intended this stretch machine as solely a gritty old school sound design tool, that makes no sense.
I’m curious if the unwanted pitch changes are just while you are actually changing the BPM, or persist after you have stopped and the BPM is static again.
So if you play at, say, 120, stop the sequencer, change BPM to 100, then hit play, do you still hear he wrong pitch?
I think it’s just because it’s a remnant of the old DT, like why they still have werp. I’m expecting a proper slice machine in an update, hence why they renamed it “grid.”
No, I haven’t tried that specifically but it’s whenever I use a loop that’s in a different BPM than the pattern that I’m working in. It’s just a lot of granular artifacts that don’t well preserve the pitch or transient integrity of the loop.
Just made a test with a loop including a continuous sine wave, recorded it and played with Stretch machine. Doesn’t seem that out of tune (a few cents, less than 5 imho) but there is a kind of tonal artifact that ruins everything, even at +/- 1 bpm. So yes, OT wins without a doubt.
Weirdly at different bpms, it reduce that tonal artifact…
As is, I’d use it as an fx only.
Not motivated to record that, barbecue time…
I noticed that too, I figured it was at certain multiples the algorithm repeats the grains at a rate that match better with the original sample or something.
Me too on OT but I will try with DT2. Tip to make timestretch slower : reduce LEN to 30 for instance to divide resulting tempo by 4, original tempo being mathematically faster with a shorter length…
With OT, if pitch corresponds to tempo (+7 semitones => tempo x 1.5, 12 semitones => tempo x 2) the sound isn’t altered at all compared to a pitched sample without timestretch. So for loop resampling to make beats from different samples at different tempo / pitch, it can be useful to change pitch to get closer to theoretical speed/pitch variation without timestretch…
Problably clear as mud, have to test…
With DT2, it seems interesting to make test with a Repitch machine comparison.
Yeah my music taste is so perverted because I expected a better time stretching algorithm from a brand new device in the year 2024 from a company that already provided a serviceable algorithm on one of their devices 15 years ago. I did not expect an algorithm that I could find in Logic, but frankly that is not much to ask for considering the fact that it is not difficult to implement with highly up to date hardware that the DT2 is platformed on, and again, the year is 2024. But I seem to not be the only one with such perversions, imagine that.