Advanced Resampling

Have an urge to learn, or a calling to teach? Want to share some useful Youtube videos? Do it here!
Post Reply
User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

25 Apr 2015

One trick you might not know about is resampling using changes in pitch combined with changes in BPM.

To do this, you can use either a regular instrument like a synth or a sampler with pre recorded audio loaded into it. Whichever device you choose, set its pitchbend range all the way up to 24. On the device's mix channel, turn on "Rec Source" in the bottom right. Make an audio track and set it to record the output of your device. Prepare whatever sequence you want in the sequencer. Now time for some math.

Take the tempo of your song project and multiply it by 4. This is the maximum value you can define in your automation curve for BPM changes. Divide your project tempo by 4 to get the minimum value for the BPM curve. (If you calculate values that are not whole numbers, adjust the minimum value to the nearest whole number and multiply to recalculate project tempo and max tempo.)

The reason we want to use 4 instead of any other number is because a change in pitch of 24 semitones (2 octaves) equals 1/4x to 4x the original playback speed.

To get this going, edit the automation of Tempo/BPM in the transport bar and edit the automation of the pitchbend for your device. Draw an automation clip into the tempo lane and fill it with a line that starts at your maximum tempo and ends at your minimum tempo. Draw another clip in the pitchbend lane and fill it with a line that starts at maximum (+8191) and ends at minimum (-8192).

Arm the audio track to record and then capture the sequence. Then, before playing it back, disable the tempo lane. You should hear phase shifts in the original sound that fade toward something similar to unison. This final sound can be exported or bounced to new recording for further mangling.

Using different slopes in combination will yield different results, but using mathematically related values for the slope values as I've shown you here tends to sound "cleaner".

This process works very nicely on whitenoise.
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

User avatar
eusti
Moderator
Posts: 2793
Joined: 15 Jan 2015

25 Apr 2015

Could you give an audio example of what is achieved with this, please?
I'm not quite getting it yet... Thanks!

D.

User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

25 Apr 2015

I will soon. The method as described above will preserve the pitch of the original sound while being recorded at different speeds. It sounds like a change in recording resolution sort of. I'll link an example when I have time.
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

User avatar
eusti
Moderator
Posts: 2793
Joined: 15 Jan 2015

25 Apr 2015

QwaizanG wrote:I will soon. The method as described above will preserve the pitch of the original sound while being recorded at different speeds. It sounds like a change in recording resolution sort of. I'll link an example when I have time.
Appreciated! Looking forward to it! :)

D.

User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

26 Apr 2015


My mistake, the process does not retain the original pitch of the sound but does automatically tempo map changes in the time domain to the project tempo.

As a minor demonstration of this effect, here are 5 sustained pulses of whitenoise that last for 4 bars at 132 bpm.
Whitenoise Example Capture

The first pulse is standard whitenoise out of Thor captured at a steady 132 bpm.
The second pulse is the result of capturing within the range of 132 to 528 bpm, resulting in harmonic distortion similar to a unison at the end of the pulse.
The third pulse exhibits the harmonic distortion that is captured when recording at a static 528 bpm.
The fourth pulse demonstrates the opposite of the second pulse with a lack of harmonic distortion when the bpm returns to 132 at the end of the tempo change.
The fifth pulse is a repeat of the standard 132 bpm capture for reference.

My original explanation wasn't very clear and I admit I was incorrect to say the pitch of the sound is retained. That is presently impossible using stock devices, and perhaps still not possible using extensions as well.

The core of the idea behind using pitchbend and tempo together is that you are adjusting the playback speed of sample-based instruments to match changes in the tempo. This nearly 1:1 relationship is established by recording the audio generated by these two changes in real time on an audio track.

If the tempo decreases, a sample playing at normal speed during recording will become compressed within a shorter space along the timeline when it is played back with the changes in tempo disabled. Pitching the instrument down to match the rate of travel of the recording head eliminates the compression.
Likewise, if the tempo increases, a sample playing at normal speed during recording will become expanded across a larger space on the timeline when it is played back without tempo changes enabled. Pitching the instrument up to match the recording speed eliminates the expansion.

A better demonstration of the effect is linked here.
Tempo Resampling Demo

The synth was captured to an audio track, then bounced to Rex and played back with a pitchbend applied to the OctoRex to match the tempo.

The first 4 bars are the original with no pitch applied.
The second 4 bars contain the result of the process using increased tempo and pitch that is being played back at the original 132 bpm.
The third 4 bars contain the result of taking the output of the second 4 bars and processing it with a downward tempo and pitch. You can hear the pitch slowly warping down and distorting the melody which is a good demonstration that the process is not truly 1:1.

As you can hear, the result is a bit odd. Typically, samples play slower when pitched down and faster when pitched up, but instead this method enables you to retain tempo while still pitching the sample. This creates a partial retention of the original harmonics of the sound but also exhibits the change in pitch that was applied, as well as harmonic distortion that is introduced by the interpolation algorithm of the stretch function.

While some say and I do somewhat agree that Reason's stretch function sounds wonky (compare Composers' Desktop Project for silky smooth audio warping), this is one way of exploiting the unique character of Reason's audio engine to make sounds that can't be generated elsewhere and are inherently unique within Reason.

This could be used to make really strange sounds, long sweeping risers, or pulsating whitenoise that needs some dynamic character.
Considering the process is not precisely 1:1 to eliminate changes in pitch, it remains an otherwise impractical effect for melodies...for now(?)
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

User avatar
Ocean of Waves
Posts: 231
Joined: 16 Jan 2015

29 Apr 2015

This I gotta do myself to understand it  :crazy:

User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

29 Apr 2015

Yeah, it's a bit hard to explain. I guess in simplest terms it's a way to pitch samples and automatically keep them synced to the tempo through the capturing process.

But if you use effects that work with tempo subdivisions or decays, you can distort those effects by ramping tempo as described above to achieve some stranger results.

I should do a video on this.
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

avasopht
Competition Winner
Posts: 3932
Joined: 16 Jan 2015

29 Apr 2015

Isn't this what Samplitude's stretch does?

User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

29 Apr 2015

From what I can tell, it does something similar. I'll have to do a proper demonstration of this I guess.
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

User avatar
Ocean of Waves
Posts: 231
Joined: 16 Jan 2015

29 Apr 2015

QwaizanG wrote: it's a way to pitch samples and automatically keep them synced to the tempo .
But it is this any different than just transpose a clip?

User avatar
Ocean of Waves
Posts: 231
Joined: 16 Jan 2015

29 Apr 2015


I've been doing a lot of this lately (@02:13), I don't use a tapemachine but NNXT


User avatar
Raveshaper
Posts: 1089
Joined: 16 Jan 2015

29 Apr 2015

Yes. It's different because you can change smoothly from one pitch to another and still keep it on the grid. Play a long sample in NN-XT or something and have it pitch down over time. The playback will slow down the more you pitch it down. This method retains the pitch but conforms it to the grid. Transposing clips can't be automated using slopes I don't think.
:reason: :ignition: :re: :refillpacker: Enhanced by DataBridge v5

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 9 guests