Electronic Music Tech Hero: David Cockerell

I was following a discussion on Chris Randall’s blog about vocoders, and the Electro-Harmonix Voice Box came up. This is a simply incredible sounding vocoder, and is “designed by the same EMS genius who made vocoding famous,” according to the EHX web site. The “EMS genius” is the mighty David Cockerell.

David Cockerell designed the EMS products* from 1969 to 1972, including the VCS3, the VCS4, the Synthi-100 (famous as the BBC Radiophonic Workshop’s “Delaware”), the Synthi/AKS, and the Hi-Fli pedal.  Cockerell then went on to design some of the classic Electro-Harmonix pedals of the 1970’s: Small Stone, 16-Second Digital Delay, the Microsynth, and so on. In the 1980’s, Cockerell designed the Akai samplers, including the S900. Today, David Cockerell is back at Electro-Harmonix, cranking out pedals such as the HOG/POG/MicroPog, the Stereo Memory Man with Hazari, the 2880 loop sampler, and the Voice Box.

As a DSP developer, I am astounded not only at the quality of the algorithms that Cockerell has developed, but also at how he got them running on fairly underpowered DSPs. The MicroPOG, for example, seems to be running an FFT-based phase vocoder algorithm with decent overlap (I would guess 4x), as I don’t hear the artifacts that can be heard when using 2x overlap. Yet the whole box runs on a Freescale DSP56364, which has plenty of cycles, but almost no internal memory. The other Electro-Harmonix digital boxes by Cockerell appear to be running on an Analog Devices BF561 Blackfin DSP, which has somewhat more memory, but is a 16 bit part. The 16 bit FFT that Analog Devices provides would be unusable for any sort of decent audio work, so the Electro-Harmonix developers must have rolled their own FFT code, presumably with higher precision and some scaling involved. The Voice Box also seems to be FFT based (the 256 bands of vocoding is a dead givaway), so similar issues have to be solved.

Searching on Google doesn’t turn up a lot of in-depth information on David Cockerell. There is an interview with Mike Matthews of EHX where Cockerell’s involvement is discussed, but I would love to find an interview with the man himself. If anyone knows of more information, please drop me a line.

* Interestingly enough, the EMS vocoders appear to have been designed by Tim Orr, not David Cockerell, Electro-Harmonix notwithstanding.

The frequency shifter and Dr. Who

Like all semi-trendy electronic music nerds, I became obsessed with the BBC Radiophonic Workshop, Delia Derbyshire, the Putney, the Delaware, and all that good stuff back around the turn of the century. At one point, my coworker Tim Stilson fired up Cool Edit 2000 (state of the art waveform editor at the time) and looked at the Dr. Who theme in the spectrogram view. The spectrogram confirmed what we had suspected: the echos at the end of the theme were shifting upwards in pitch.

I soon found out that these shifting echos were the product of a frequency shifter with a tape echo in its feedback loop. Apparently this combination was a popular effect in Delia Derbyshire’s toolkit, and was used most notably to create the electronic scream that was played before the end credits:

The following description is from Mark Ayres’ most excellent history of the Doctor Who theme:

The sound consists of two elements, a rising bubbling sound and a descending scream effect. First of all, let’s deal with the rising bubbling sound. The process used to create this was very simple. The first couple of notes of the closing titles (as the theme melody enters) were copied onto a new piece of audio tape. This was leadered, then flipped over so that it played backwards. The output of the tape deck was then fed into a frequency shifter set for a downwards shift at a short delay, and fed back into itself. When the tape was played into the shifter, it came out the other end milliseconds later with its pitch shifted down slightly. This output was fed back to the input and so on, creating a downwards cascade of ever more distorted sound. This was copied onto a new tape, and when this piece of tape was in turn flipped over, Delia was left with a rising flood of sound starting very distorted and slowly resolving into the opening couple of notes – this was then spliced onto the beginning of the theme.

The downwards scream was created in similar fashion. The source sound is a downwards-sliding hard-edged tone produced using an audio oscillator. Again, this was fed into the pitch shifter with very short delay, a downwards shift, and heavy feedback. Aliasing distortion within the shifter added to the overall effect. The result was mixed with the rising echoes to give the sound we are all familiar with.

I have been working on recreating these sounds with my VST frequency shifter / echo plugin, with some degrees of success. The overdrive in my analog echo simulator sounds a bit soft, but the harsh edged “scream” is also dependent on having a more edgy tone being fed into it. The process would be easier if I exposed tone controls for the internal filters, so I think that will be my next step. I find myself enjoying the sounds that the plugin makes on its own, without emulating any particular existing sounds. However, the Dr. Who sounds are so archetypal, it would be good to have a tool that can make similar sounds, so I will keep working towards that.

UPDATE, April 2016: Since writing the above in 2009, I have released an “official” plugin that incorporates analog echo and a frequency shifter in a feedback loop, ValhallaFreqEcho. It’s free, so go ahead and download it!

Frequency Shifting, 10 years later

Back in the spring of 1999, I was taking a yearlong computer music course at the University of Washington. I wasn’t a matriculated student at the time. I talked my way into the course, as I was obsessed with getting the sounds in my head out into the world.

One of my obsessions at the time was creating a digital emulation of the Bode Frequency Shifter. A frequency shifter is basically one half of a ring modulator. While a ring modulator takes an input signal and a modulation signal, and produces the sum and difference frequencies, a frequency shifter will produce just the sum frequencies, or just the difference frequencies. It turns out that separating the frequencies is actually fairly complicated, and involves some fancy filtering, as well as working with complex numbers. I hadn’t dealt with complex numbers since high school, so I rolled up my sleves and set to work.

After a lot of dead ends and swearing, I created a high quality Hilbert network, which is the heart of the Bode frequency shifter. My Hilbert network had all of the nice artifacts of the original analog phase differencing network, which allowed me to get the barberpole phasing sounds I always wanted to hear. I implemented the Hilbert filter in Csound, which was very powerful for the time, but seems amazingly primitive to me today. My code would take several minutes to render, but when I heard the results, I was a happy man.

A lot of experimentation followed. I put the frequency shifter in feedback loops, using delay lines and other processing blocks, to get Risset endless glissandos out of any source material. I used the frequency shifter to FM input signals, and fed the output back to the modulation inputs, for some weird phase locked loop tricks (I need to track that work down again). Finally, I tried out my newly acquired C skills, and implemented the Hilbert network as a Csound unit generator.

This spring marks 10 years that I have been working with frequency shifting. I have implemented versions of the frequency shifter for a lot of platforms, from Reaktor and Supercollider to the original Xbox audio DSP.  After all these years, I have finally created a version of my frequency shifter code in VST form (with AU code to follow). You can get it here. I have added some new tricks: the frequency shifter is in series with an analog echo emulation. Turn up the feedback, and you can get screeching endless glissandos, barberpole flanging, or soft echos that rotate around and through your skull. It’s a work in progress, but it sounds pretty nice right now.

An introduction

Welcome to the Halls of Valhalla! 

A bit of background: For the past 20+ years, I have been interested in the intersection between the academic world, the professional world, and musicians. I am fascinated by electronic and computer music on many levels:

  • Technical: I am an audio DSP developer by trade, so I like to dive into the nitty-gritty of how things work. 
  • Musical: My tastes tend more towards the popular than the classical/academic, but I am fascinated on how these worlds have overlapped over time.
  • Historical: I love following the life cycle of ideas, from when it was first developed, to how it was used by musicians, to when it was declared obsolete, only to rise phoenix-like from pawnshops, old journals, and so on.

The purpose of this blog is to explore all of the aspects of the intersection between technology and music that I find interesting at any given moment. I will probably have an emphasis on the technical aspects of things, but tempered with discussions on how technical ideas make their way into popular music. Things will probably be a bit scattershot at first, but over time the random dots should connect together to form a beautiful pattern that explains all audio things to everyone on earth.