In our series Pile of vintage brochures we now present you the EMS Vocoder.
Below you will find:
- Scans of the original (UK) brochure, dating from (approx) 1970
- Specifications (taken from the last page of the brochure
- A downloadable pdf of this brochure
- A link to a youtube video demonstrating the EMS vocoder capabilities.
The EMS-VOCODER produces, modifies and treats speech and other sounds in new and imaginative ways. For example, it can
- Synthesize speech at constant pitch
- Synthesize speech at varying pitch
- Transmute the apparent age or sex of speaker-or change a single voice iRtO a chorus or make a voice speak with unnatural depth or-height
- Articulate speech on to another sound, such as train or car engine noises or musical sounds or natural sounds such as wind or water. Particularly it can articulate human choruses like a football crowd. One musical instrument can articulate another.
- Change the rate of speech independent of the pitch
- Freeze the spectrum of a spoken word
Clarity of articulation is excellent and the technical quality (signal to noise ratio of unwanted distortion, etc.) is comparable to existing audio standards.
The EMS-VOCODER works by analysing a speech signal into frequency bands. The energy from each band is then converted into voltage which controls the gain of a synthesis filter bank. The synthesis section is another bank of filters covering the same frequency band as that of the analyser. An excitation is fed into the synthesis filter and the speech spectrum extracted from the analysing filters is imposed upon it.
Other devices determine whether the speech is voiced or unvoiced and also the fundamental pitch of the speech.
The EMS-VOCODER can be computer monitored and controlled and forms a natural extension to the EMS Computer Synthi and Synthi 100 suite of equipment.
VOCODER TECHNICAL DESCRIPTION
SPEECH AND EXTERNAL EXCITATION INPUTS:
Input impedance: 10K unbalanced Signal level: -20dBm min.)
VOCODER OUTPUT: Output impedance: low – will drive 600 ohms at. line level Signal level: 0dBm unbalanced
REAL TIME SPECTRUM DISPLAY:
Outputs via two 75 ohm BNC connectors
Oscilloscope requirements XY mode 0.5v/cm sensitivity.
SIGNAL LEVEL MONITORING: Speech, excitation and Vocoder output levels are monitored on three PPM meters.
ANALYSING AND SYNTHESIZING FILTER BANK:
The filter bank is made up of 22 channels each containing an analysing filter, an envelope follower, a slew limiter, a voltage controlled amplifier and a synthesizing filter.
Twenty of the filters are 8-pole bandpass units having a quarter octave spacing with centre frequencies of 205, 249 303, 367, 444, 539, 653, 791, 958, 1161, 1406, 1703, 2064, 2500, 3030, 3670, 4447, 5388, 7888 Hz (3% tolerance). The remaining two filters are a high pass and low pass at either end of the band pass range. The envelope followers have a low pass smoothing proportional to the centre frequency of the filter and a dynamic range of 54dB
The bandpass filters have an ultimate roll off of 24d8/octave. Adjacent channel rejection is 16dB at the neighbouring centre frequency. The filter signal to noise ratio is 72dB.
Speech and excitation signal breakthrough in the filter bank is minus 50 and 60 dB respectively. This is reduced to minus 70dB with the squelch function switch. Each filter produces an envelope output and has a control voltage input to externally control the synthesizing channel gain. All 22 outputs and 22 inputs are avaiIable on a large mu ltiway connector at the rear of the machine (computer interface).
Envelope outputs: 0 to -4v linear with respect to signal levels
Control inputs: 0 to + 4v linear with respect to signal levels. Signals from the envelope followers to the slew limiter/freeze section droop in freeze mode less than 0.1d8/sec.
PITCH EXTRACTOR: This produces a voltage proportional to the logarithm of the frequency of the glottal pulses in the speech input.
Frequency range: 50 – 1000 Hz
Response time for 6% accuracy: 5 cycles or 20ms (the greater).
Output voltage: + 0.5v/octave and + 1.1 to -1.1v/octave. Input low pass filter (qua I ity control): 100 – 1000 Hz.
Overtone rejection: 20dB (fundamental must have at least 3% of the total energy to achieve the stated accuracy).
The OUTPUT VOLTAGE is available at the rear panel.
PATCHING SYSTEM: 22 x 22 patchboard,using shortin~ pins.
EXCITATION SOURCES: 2 EXTERNAL
EXCITATION INPUTS: Input impedance: 10K unbalanced
Signal level: -20dB (Minimum).
These signals may be gated by the VOICED or the UNVOICED decision Switches or may be left ‘ON’ continuously.
INTERNAL NOISE SOURCE: Controls include gating, level and ‘colour’ by a variable frequency, low Q, low pass filter.
2 INTERNAL VOLTAGE CONTROLLED OSCILLATORS:
Output waveforms: Square/ramp.
Pitch spread (from an external voltage): + 0.5v octave.
Controls include gating, frequency, level, pitch extraction, external voltage and keyboard input.
This unit has no external controls, but provides gating signals for the external excitation, noise source and voltage control led osci I lators. A green lamp indicates a voiced decision, an orange lamp an unvoiced decision.
This may be manually operated with the variable slew control or a freeze switch as well as with an external voltage control. Voltage control: 0v to + 1v (Freeze).
Range: + /-0.05Hz to 1000Hz (20,000:1 range).
Carrier breakthrough less than 60d8.
OUTPUTS include up-shifted, downshifted, original and mixed versions. KEYBOARD CONTROLLER: This unit is designed to interface the voltage controlled oscillators with the standard voltages from EMS keyboards.
POWER REQUIREMENTS: Voltage: 220v AC or 110v AC. Switchable. Power: 80 watts. Maxiimum.
SIGNAL CONNECTORS: These are all . inch mono-jack connectors, except for the spectrum display (BNC) and the computer interface.
MECHANICAL DETAILS: Maximum dimensions: 35″ x 9 ½ ” x 21 ½” (90 x 24 x 55 cm).
Weight: Approximately 45 lb./20 kg.
Click the image on the right to watch the demonstration video.