[sdiy] Formant filters, yay!

Tom Wiltshire tom at electricdruid.net
Fri Mar 10 19:50:08 CET 2017


If you get to the point that people can tell what accent your vowel filter has, you've nailed it!

You'll find huge differences in vowels between speakers of many languages, but that's no reason not to give it a go. So what if my module sounds more Newcastle than London?! Get it vocal-sounding and then tweak to taste. After all, there are significant differences in formant frequencies between male and female speakers, or between adults and children, but we all mange to be understood well enough (mostly).

I like the idea of using the phonetic alphabet as a reference point though - so it wouldn't just be an "a", but a specific "a" sound.


On 10 Mar 2017, at 17:30, Dave Garfield <daveogarf at yahoo.com> wrote:

> Very good point, Florian!
> There is some ambiguity in pronunciation among languages.  Would using the phonetic alphabet, where pronunciation is spelled out precisely, help to clear things up? 
> In English, vowels like "a" and "I" and "u" are diphthongs, consisting of more than one sound combined.  This would involve creating more than one sound, and morphing between them.
> Ah, yes.  What a tangled web we have in language!
> All the best,
> Dave Garfield - Colorado, USA
> 
> 
> On Friday, March 10, 2017 2:40 AM, Elaine Klopke <functionofform at gmail.com> wrote:
> 
> 
> So what would you suggest, Florian? My assumption was that this was a rough guideline to start from. Obviously any filters that I make aren't going to be perfect and this particular table of stats was the first thing that I came across... 
> 
> Is there a more accepted list of frequencies and such to go off of?
> 
> Sent from my iPhone
> 
> > On Mar 10, 2017, at 2:59 AM, Florian Anwander <fanwander at mnet-online.de> wrote:
> > 
> > And just something more "fishy" about synthesizing vowels or vocoding:
> > 
> > What does mean "a" or  "o"?
> > "A" in which language Japanese or English? Which dialect: Edinborough or London? Which local slang: North or south of Wandsworth Road?
> > 
> > I think vocoders or voice imitating filters will sound very different, because they were developed  by engineers, who speak different languages and who are used completely different vocalisations. If an english person pronounces an "o", it sounds to german ears more like an (german) open "u". Dieter Doepfer is born in Frankonia and speaks his "a" more open. The Dynacord engineers sit in Lower Bavaria, their "a" is dark and tends to be an "o".  How sounds the A-129 compared to a SRV-66? Ok, I know the SRV-66 is originally a dutch syntovox 222. But I think you get the principle:
> > 
> > If an engineer tries to optimize a vocoder for vowels, or tries to create a vocal filter, he will optimize his design for his imagination of the vowel, and his device will sound in the vocalisation of his language fine, but will fail with some other vocalisations.
> > 
> > Florian
> > _______________________________________________
> > Synth-diy mailing list
> > Synth-diy at synth-diy.org
> > http://synth-diy.org/mailman/listinfo/synth-diy
> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20170310/9cb791ba/attachment.htm>


More information about the Synth-diy mailing list