[sdiy] MXR Pitch Transposer Schem/Circuit

Martin Fay martin at martinfay.com
Mon Mar 11 10:55:19 CET 2013

Forgot to copy the list...

> Hi Richie,
> Sean's blog has a whole series of posts on pitchshifting covering  
> tape-based onward. My description more or less matches the H910  
> which used triangle waves for the crossfade.
> I think something like this could be done with BBDs but the  
> artifacts of the modulated clock would surely be "interesting"!
> Martin
> Sent from my iPhone
> On 10 Mar 2013, at 20:21, Richie Burnett  
> <rburnett at richieburnett.co.uk> wrote:
>> The two read-pointers with cross-fading to minimise the glitches is  
>> really the bare minimum for pitch shifting. It still sounds quite  
>> rough though in practice. Even though you fade the contribution of  
>> each read pointer down to zero when it crosses the write pointer  
>> theres still a problem unless a whole number of audio cycles fit  
>> perfectly into the ring buffer's memory. Essentially the waveforms  
>> from the two read pointers are rarely coherent for arbitrary audio  
>> inputs. So when you cross-fade between them you get constructive/ 
>> destructive interference that causes wobbles in the amplitude and  
>> phase of the pitch shifted audio output. This sounds rough.
>> Any decent pro-audio pitch shifter these days also performs an auto- 
>> correlation on the incoming audio to determine any periodicity.  
>> Then each of the two read pointers are moved by whole numbers of  
>> audio periods when they would cross the write pointer and otherwise  
>> cause a glitch. If you nudge the read pointers by whole numbers of  
>> audio cycles then there should be no glitches in theory. However  
>> cross-fading is still used to minimise any glitches caused by less  
>> than perfect period detection or poor auto-correlation when pitch- 
>> shifting complex polyphonic material. At least now the waveforms  
>> from the two playback pointers are as coherent as they can be.  
>> Almost identical and in-phase. So the cross-fading now causes  
>> minimal destructive interference and amplitude wobble.
>> That's how pitch shifting has been done since Eventide's H949  
>> Harmonizer in 1977. Sean Costello provides a good explanation here:
>> http://valhalladsp.wordpress.com/2010/05/07/pitch-shifting-the-h949-and-de-glitching/
>> -Richie,
>> Sent from my Sony Ericsson Xperia ray
>> Martin Fay <martin at martinfay.com> wrote:
>>> Hi Gordon,
>>> I believe the trick is to have 2 read pointers into the delay buffer
>>> and crossfade between those outputs to avoid reading close to the  
>>> saw
>>> reset.
>>> Martin
>>> Sent from my iPhone
>>> On 10 Mar 2013, at 10:00, Gordon JC Pearce <gordonjcp at gjcp.net>  
>>> wrote:
>>>> On 08/03/13 16:13, Tom Wiltshire wrote:
>>>>> That's a damn interesting gizmo. Does anyone have any idea how it
>>>>> works?
>>>> Someone on the SQ80 mailing list - where we were discussing time-
>>>> based effects the other day, as it happens - mentioned a pitch
>>>> shifter that modulated the delay clock with a sawtooth.  I can see
>>>> how that would give a steady-ish pitch shift but surely it would
>>>> "hiccup" as the saw reset?
>>>> -- 
>>>> Gordonjcp MM0YEQ
>>>> _______________________________________________
>>>> Synth-diy mailing list
>>>> Synth-diy at dropmix.xs4all.nl
>>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at dropmix.xs4all.nl
>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy

More information about the Synth-diy mailing list