[sdiy] Hybrid (analog/digital) vocoder?

[cheater cheater]

The digital domain will be much better for level detection of lower
bands, and depending on how you realize it, the higher parts too. Of
course you would need a separate audio ADC for each vocoder band. But
one such chip is simpler than a whole level detector, and you can in
that case also use different metering algorithms, which can change the
character of the vocoder substantially. The signal can bypass the
metering block and stay analog this way; the vca's shouldn't be
difficult, simple multiplying DAC application.

If you go with separate ADCs then there will be no problem with 'models' etc.

However as I understand it - correct me if I'm wrong - you need a
separate bank of filters for the carrier and the modulator anyway, and
the modulator would probably be much simpler, therefore different, so
no reason not to do it in digital.

D.

On Mon, Feb 15, 2010 at 09:40, Naoki Iwakami <naoki.iwakami at gmail.com> wrote:
> Hello John,
>
> I'm not sure which of analog or digital processing is superior for a vocoder.
> Mostly I agree with you but there should be a lot of other opinions here.
> But when you are working on analog and want an analog VCA, I think the
> PWM type has some advantages against traditional OTA type VCA.
> PWM VCA is simple and linear.
>
> You can create PWM using micro processor but actually you can also obtain it
> using a triangular generator and a comparator, just as usual VCO does.
> If the VCAs are multi-channel, they can share the trianglar generator.
> So you just need to add a comparator and a switch to increase a VCA channel.
> Roland VP-330 uses a single transistor for a comparator.  So its VCAs are quite
> simple compared to OTA-type.  I thinks this is an advantage in "analog domain".
>
> Also, the PWM type is quite linear as long as the switch is fast enough and
> the rectangular is linear enough.  It is hard to obtain for a
> transconductance type
> VCA with such a small set of parts.
>
> So other than vocoder, the technique might be interesting to be used for some
> analog application like multi-channel compressor or a mixer.
>
> Thank you,
> -- Gan
>
>
> On Mon, Feb 15, 2010 at 1:35 PM, John Mahoney <jmahoney at gate.net> wrote:
>> At 10:56 PM 2/14/2010, Paul Perry wrote:
>>>
>>> Just saw mention in the CMOS switch as VCA thread that some famous
>>> vocoders use PWM for the VCA.
>>> So I thought - not to complicate matters - to start a new thread, on the
>>> question of  a 'hybrid' vocoder, in which the  filters are analog, but the
>>> VCAs are digital, and the PWM is controlled by DSP.
>>> The idea is, the carrier remains analog throughout, with a PWM VCA.
>>> But the analysing is done in a DSP or micro, digital outputs from which
>>> control the various filters in a PWM fashion.
>>>
>>> This would have the advantage that
>>> 1. the filters are all analog, giving possiblility of resonance knobs on
>>> each.
>>> 2, the PWM - which can be tricky - can do the difficult parts - no
>>> comparators needed - plus the analysing filtering can be in software
>>> without affecting 'musical' quality.
>>
>> Interesting.
>>
>> Regarding point #1, digital filters can also have resonance knobs, can't
>> they?
>>
>> A benefit of analog circuitry is easier access to each band's output, for
>> special effects purposes. Then again, multiple-output DACs are not so
>> expensive, anymore, so maybe this is a wash.
>>
>> The analysis (digital) filters should have the same characteristics as the
>> analog ones, so a bit of profiling and model tweaking may need to be done, I
>> guess.
>>
>> Analog vocoders have far fewer bands than digital ones. More bands sound
>> more accurate. Seems that using analog circuits is the limiting factor,
>> presumably because of cost and/or size. Are the benefits of "pure analog
>> sound" worth the trade-offs?
>>
>> There are some pretty good analog filter models, these days, as well as
>> reasonably convincing models of tube circuits and other analog components.
>> Therefore, using DSP doesn't have to mean that the sound has no character.
>> In fact, a DSP-based unit can offer a variety of models and modes. "Bode
>> mode," varieties of clean and dirty sounds, reverse mapping, formant
>> shifting, etc.
>>
>> I'm thinking that it's a neat idea but not worth doing. Just my opinion.
>>
>> Curious to see more comments,
>> John
>>
>> _______________________________________________
>> 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
>