I've heard this task is possible albeit difficult, but I was wondering how I might go about building an analog pitch shifter. I have a few conceptual ideas, but I like the electronics knowledge to start them and I was wondering if anyone could help. Here are two different ways I've considered pitch shifting:
1) Split an audio signal in two. Take one of the signals and run it through an oscillator so it starts at the right pitch, and then allow whatever control to bend the pitch of the oscillator. Then, add the two signals together. This isn't really efficient or effective pitch shifting since the original signal has timbre but lacks the bent pitch, and the bent pitch lacks timbre. This is just an effect, and may or may not sound good.
2) This idea is the one I really want to pursue because I think it has more potential. I was reading about sampling a signal and how you have to sample at 2 times the highest frequency you're recording inorder for it to sound right. If you drop your sampling rate below that, you get aliasing witch maintains the timbre of the signal, but alters the pitch. I was wondering if you could somehow manipulate this to achieve pitch bending
Any help would be greatly appreciated. I need to know what circuits would help me achieve this, schematics would be great. Thanks in advance!
1) Split an audio signal in two. Take one of the signals and run it through an oscillator so it starts at the right pitch, and then allow whatever control to bend the pitch of the oscillator. Then, add the two signals together. This isn't really efficient or effective pitch shifting since the original signal has timbre but lacks the bent pitch, and the bent pitch lacks timbre. This is just an effect, and may or may not sound good.
2) This idea is the one I really want to pursue because I think it has more potential. I was reading about sampling a signal and how you have to sample at 2 times the highest frequency you're recording inorder for it to sound right. If you drop your sampling rate below that, you get aliasing witch maintains the timbre of the signal, but alters the pitch. I was wondering if you could somehow manipulate this to achieve pitch bending
Any help would be greatly appreciated. I need to know what circuits would help me achieve this, schematics would be great. Thanks in advance!
Analogue pitch shifting is done by modulating the signal with one carrier frequency but demodulating it with a slightly different carrier frequency. The famous example of this is the Dalek voice in Dr Who. You require two very stable carrier oscillators and two four-quadrant multipliers. No, I don't have a diagram.
What do you exactly want to do ? Do you really want to SHIFT frequency ?
Shifting means as mentioned by EC8010 that it is incresed or decreased by a fixed amount like making 315 Hz out of 300, 415 out of 400, 515 out of 500 ..........
Or do you want to multiply or divide ? And in this case it makes a huge difference if it is just a single frequency or a whole frequency band and if it is distortion-sensitive.
Regards
Charles
Shifting means as mentioned by EC8010 that it is incresed or decreased by a fixed amount like making 315 Hz out of 300, 415 out of 400, 515 out of 500 ..........
Or do you want to multiply or divide ? And in this case it makes a huge difference if it is just a single frequency or a whole frequency band and if it is distortion-sensitive.
Regards
Charles
I basically want to make an electronic analog whammy bar. It doesn't have to have an actual "whammy bar" but I want to be able to bend notes full octaves on command with a potentiometer or something. Any ideas?
Oh, and I was also thinking of a third way to be able to bend notes non-digitally:
Take the incoming signal, split it up into discreet "packets" and speed up or slow down each packet uniformly. Changing the speed changes the tone, then you'd put each packet back together and its pitch would be changed.
Oh, and I was also thinking of a third way to be able to bend notes non-digitally:
Take the incoming signal, split it up into discreet "packets" and speed up or slow down each packet uniformly. Changing the speed changes the tone, then you'd put each packet back together and its pitch would be changed.
You might try to find an old Radio Shack reverb box that used the analog bucket brigade chip. It was an analog level cmos shift register I.C. You change the frequency of the clock oscillator with a potentiometer and it will momentarily change the speed and thus pitch of the audio that tumbles out of the back end of the brigade chip. This is exactly the methodology you postulated in your previous post, but in an analog domain (audio not converted to digital signal). The effect is actually like seriously bad wow and flutter in a tape deck.
The box has a few slide pots and uses a 9 volt battery. Line level inputs and outputs. RCA jacks.
I have one still in the original box (somewhere) that could be pried loose if you cannot find something closer. Contact me off list if you want it.
The box has a few slide pots and uses a 9 volt battery. Line level inputs and outputs. RCA jacks.
I have one still in the original box (somewhere) that could be pried loose if you cannot find something closer. Contact me off list if you want it.
If you're interested, there are currently some Radio Shack Reverbs on eBay going for ~$10 - $15 each:
http://search.ebay.com/search/searc...from=R8&satitle=radio+shack+reverb&category0=
- Rickshaw
http://search.ebay.com/search/searc...from=R8&satitle=radio+shack+reverb&category0=
- Rickshaw
I think EC8010's probably got the best idea for this.
I used to have an original Digitech Whammy pitch shifting pedal that costs a fortune.
They, and the newer ones, use digital sampling for the shifting I think. The originals were supposidely better because the shifting chip was produced by a company who specialised in custom firmware, whereas when it came round to making the new version Digitech decided to do it themselves and didn't do so well.
I've never tried one of the new versions, so I can't comment. But the original shifter I had was very slick and responsive. It's a tricky effect to subtly fit into a sound though, it's pretty in your face. The octave up sound had a kind of ring modulated, tinny, phased, flanged, chorusy side to it if you actually tried playing long passages through it. You can hear it on some of Audioslaves stuff - I think one tune is 'Like a Stone', but I can't remember for sure. The harmoniser effects could produce some extremely thick sounds. Like in the break in Cochise, also by Audioslave. I like the bit in Shadow on the Sun when Morello does the huge octave up sweeps with all the delay and funny noodling, I thought that was pretty innovative. I really like the sound in that solo. But the high pitched whistling effects get very tedious in a lot of their stuff and the tunes he did with RATM.
I'm guessing the original poster is asking about analog shifting because a lot of the digital shifters you can get now sound nasty. I expect he also wants a solder iron option, as opposed to coding.
I'm kind of interested in this as well, but as far as I know, you're not going to find a schematic for something EC8010 described for the studio in hurry, all of the variable shifters I've seen have been digital. Someone was asking about how to code a digital shifter on here I think, and I don't remember it coming out as easy. Besides, you'd probably just want to buy one if you go that way, unless you're already a coding genius.
Like Neutron says, pitch shifting effects like the Octavia rely on a blisteringly simple method of frequency doubling that will only double the frequency, you can't make it variable.
To go any further, I'd suggest reading around to see if you can find examples of the Darlek voice in Dr Who. You'll also find mountains of stuff on modulation with carriers by reading about radio electronics. Expect to read quite a lot is the only advice I can give at the moment about that. Introductions to basic radio should start discussing carriers early on. A lot of it is suprisingly easy to understand and interesting, for me anyway.
There's a site on the net somewhere called something like 'Radio school' with a load of PDFs on it about how radio works for the beginner. I remember that being an excellent read but can't find it again.
One good thing working in your favour here is that there are also tons of premade ICs for radio electronics that you might be able to hack into a pitch shifter. You might even be able to bust something out of radios, vcrs and tvs depending on how inventive you are.
I used to have an original Digitech Whammy pitch shifting pedal that costs a fortune.
They, and the newer ones, use digital sampling for the shifting I think. The originals were supposidely better because the shifting chip was produced by a company who specialised in custom firmware, whereas when it came round to making the new version Digitech decided to do it themselves and didn't do so well.
I've never tried one of the new versions, so I can't comment. But the original shifter I had was very slick and responsive. It's a tricky effect to subtly fit into a sound though, it's pretty in your face. The octave up sound had a kind of ring modulated, tinny, phased, flanged, chorusy side to it if you actually tried playing long passages through it. You can hear it on some of Audioslaves stuff - I think one tune is 'Like a Stone', but I can't remember for sure. The harmoniser effects could produce some extremely thick sounds. Like in the break in Cochise, also by Audioslave. I like the bit in Shadow on the Sun when Morello does the huge octave up sweeps with all the delay and funny noodling, I thought that was pretty innovative. I really like the sound in that solo. But the high pitched whistling effects get very tedious in a lot of their stuff and the tunes he did with RATM.
I'm guessing the original poster is asking about analog shifting because a lot of the digital shifters you can get now sound nasty. I expect he also wants a solder iron option, as opposed to coding.
I'm kind of interested in this as well, but as far as I know, you're not going to find a schematic for something EC8010 described for the studio in hurry, all of the variable shifters I've seen have been digital. Someone was asking about how to code a digital shifter on here I think, and I don't remember it coming out as easy. Besides, you'd probably just want to buy one if you go that way, unless you're already a coding genius.
Like Neutron says, pitch shifting effects like the Octavia rely on a blisteringly simple method of frequency doubling that will only double the frequency, you can't make it variable.
To go any further, I'd suggest reading around to see if you can find examples of the Darlek voice in Dr Who. You'll also find mountains of stuff on modulation with carriers by reading about radio electronics. Expect to read quite a lot is the only advice I can give at the moment about that. Introductions to basic radio should start discussing carriers early on. A lot of it is suprisingly easy to understand and interesting, for me anyway.
There's a site on the net somewhere called something like 'Radio school' with a load of PDFs on it about how radio works for the beginner. I remember that being an excellent read but can't find it again.
One good thing working in your favour here is that there are also tons of premade ICs for radio electronics that you might be able to hack into a pitch shifter. You might even be able to bust something out of radios, vcrs and tvs depending on how inventive you are.
That is FREQUENCY shifting, not pitch shifting. You are describing the "moog frequency shifter", invented by Dr. Harald Bode.
I have read somewhere that there existed a professional analog machine for pitch shifting without changing tempo, that used some kind of magnetic media that was read in a funny way. Something like a disc that spinned at some speed while a rotating head read a specific section of the disc. So if we wanted to double the pitch every X length part of the audio was read twice at twice the speed while the disc rotated at the "correct" speed.
NOTE: I did not understand how that latter device worked, i'm just guessing.
According to wikipedia, that is just ring modulation (multiplying the audio frequency with an oscillator's out.)
What do you want to shift? Voice?
You can sort-of pitch shift voice if you run it through a vocoder and replace the voice frequency with an oscillator frequency set to double/triple/etc. of the original frequency. For that you would need a fundamental frequency analog detector, which is another problem on itself (but still solvable.)
Then you apply the vocoder on that oscillator signal, the result is a voice that is double frequency... and sounds very strange...
I realize this is an old dead thread, but this might be helpful for anyone still interested in this project. I was considering it last year through the use of BBD chips, but I suck at electronics and it seems like a large project even for a pro. But, it totally works in my head and is based on the only analog pitch shifter I ever heard about that was tape-based.
If I understand correctly it used a tape loop. Within the tape loop, there is an arc of tape that goes over a wheel fitted with multiple play heads. Signal is recorded to the tape with a stationary head just before the tape head wheel, and then the tape heads can stay stationary for normal playback, or spin faster or slower than the arc of tape to change pitch. Constant tape head wheel speed equals a stable interval of the input. Accelerate or decelerate for a gliding pitch change.
There's a Foreigner song that uses that kind of pitch shifter to generate a higher vocal harmony, and I had the chance to hear the isolated vocal tracks. You don't really notice it in the full mix, but the pitch shifted track has a wobble to it because the spinning playback heads overlap bits of the original signal, and probably don't crossfade gracefully.
I figure that a similar effect might work using BBD modulated delays, but it would be tricky to get the timing right. Each delay line (and you would need several) would need a ramp waveform LFO to control the delay time from low to high or high to low depending on the desired pitch shift, and they'd have to be slightly offset from each other. You'd also need to build an auto mixer to crossfade between each delay line tied to the same clock. The delay lines work round-robin. I'm tired so I'm not thinking through how to set up adjusting the pitch range, but I think it's all possible.
The more delay lines and closer the timing, the smoother the result. So to make it sound good, I think it'd have to be pretty expensive parts-wise, take a long time, and end up pretty large.
If anyone's crazy enough to try it, I'd love to see/hear the result!
-Chris
If I understand correctly it used a tape loop. Within the tape loop, there is an arc of tape that goes over a wheel fitted with multiple play heads. Signal is recorded to the tape with a stationary head just before the tape head wheel, and then the tape heads can stay stationary for normal playback, or spin faster or slower than the arc of tape to change pitch. Constant tape head wheel speed equals a stable interval of the input. Accelerate or decelerate for a gliding pitch change.
There's a Foreigner song that uses that kind of pitch shifter to generate a higher vocal harmony, and I had the chance to hear the isolated vocal tracks. You don't really notice it in the full mix, but the pitch shifted track has a wobble to it because the spinning playback heads overlap bits of the original signal, and probably don't crossfade gracefully.
I figure that a similar effect might work using BBD modulated delays, but it would be tricky to get the timing right. Each delay line (and you would need several) would need a ramp waveform LFO to control the delay time from low to high or high to low depending on the desired pitch shift, and they'd have to be slightly offset from each other. You'd also need to build an auto mixer to crossfade between each delay line tied to the same clock. The delay lines work round-robin. I'm tired so I'm not thinking through how to set up adjusting the pitch range, but I think it's all possible.
The more delay lines and closer the timing, the smoother the result. So to make it sound good, I think it'd have to be pretty expensive parts-wise, take a long time, and end up pretty large.
If anyone's crazy enough to try it, I'd love to see/hear the result!
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.