Tag Archives: FFT

Direct Synthesis – continued

Looking at the FFT of a vowel, you can see that each formant has it’s own frequency and amplitude. These show up as “blobs” – some stretching for the duration of the wave, others for only a portion: For example, … Continue reading

Posted in Development | Tagged , , | Leave a comment

Editing Formants

I’ve mentioned before that one of the features of Praat is that it can calculate formants. (Happy dance) However, testing has shown that data can’t be used “as is” – there’s too much noise and error. That means that the … Continue reading

Posted in Development | Tagged , , , , , , , , , | Leave a comment

Praat is Awesome

The folks who had created S.A.M. (Software Automatic Mouth) some 30 years ago have a company called SoftVoice. There are a number of audio demos on their site, including the song Twinkle Twinkle, Little Star. It’s impressive enough for me to think this might be … Continue reading

Posted in Development | Tagged , , , , , , , | 4 Comments

Analyzing Audio

The only thing more tedious than recording the bazillion audio files needed to represent English speech is the segmenting that data – for example telling synSinger where in the audio sample “ca” the consonant “c” ends and the vowel “a” begins. So naturally, my … Continue reading

Posted in Development | Tagged , , , , , , , , | Leave a comment

Finally, Singing Again

I corrected a number of bugs in the positioning code, so synSinger correctly positions syllable elements. I move the PSOLA code into a static class, and incorporated it into the rendering engine – basically replacing a single call – and … Continue reading

Posted in Development | Tagged , , , , | Leave a comment

Still Messing with PSOLA

PSOLA is a real fiddly sort of beast – get one thing wrong, and audio becomes phasey and clicky. You can’t really see the problem, because the waves look good – the issue may be that they’re off by just … Continue reading

Posted in Development | Tagged , , , , , | Leave a comment

Grumble…

I’ve been trying to implement spectral morphing, but have run into a lot of problems. The morphing itself works pretty well – I can apply an envelope taken from one sound, and apply it to another with a fair amount … Continue reading

Posted in Uncategorized | Tagged , , , , , , , | Leave a comment