login

online now

We have 110 guests and no members online

search

 
Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1

TOPIC: Curiosity about voice synthesizer

Curiosity about voice synthesizer 1 year 9 months ago #830

I first found GinSing when trying to make a voice synthesizer of my own. The approach I had was first getting a clean recording of all the different phonemes for English. Then I went into Audacity to cut out all the white noise and isolate each individual sound. Then I set my Arduino Uno to do a analogRead of my computer's headphone jack(after I ran the output through a clamper circuit, so all the voltages were positive), which it then sent directly to the serial monitor. On my breadboard, I had a resistor/transistor ladder that allowed my Arduino to output signals 0-5v at 0.5v increments, and all of the signals fed into a single speaker. From the data obtained, I looked at the numbers to decide the closest 0.5v increment they should go to. I then tried to write functions for each of the different phonemes, matching the output voltage of the Arduino to vary just the same as what I read from the initial recording. Took many hours, and at the end of the day, it didn't even remotely sound like the simply word I was trying to do(Bradley, my name). I'm just curious as to the approach you took. I tried to dissect the source code, but as someone who isn't the greatest at coding, I was struggling to find exactly how the phonemes are created.
The administrator has disabled public write access.
The following user(s) said Thank You: administrator

Re: Curiosity about voice synthesizer 1 year 9 months ago #831

That's a great way to give phonetic reconstruction a shot at speech. The GinSing does is in a quite a different way that does NOT involve using pre-recorded samples. There are several advantages to this method, including the ability to change pitch, timbre, speed, inflection and so on.

The Babblebot IC is actually a waveform synthesizer that is built in such a way as to emulate the physics of the human voice using vocal tract and voice box. The "ooh" sound for example can be thought of as a pure tone that is filtered and modulated with specific parameters of a synthesizer. The the Babblebot IC does is contain a set of parameters for each phoneme (or allophone in this case) that switch (blend) based on the voice fragments being played over time. The interface then is just setting which allophone to play, along with some dynamic parameters such as voice pitch and so on. You can find out more if you check out the Babblebot IC data sheet link on the front page.

If you have any questions, please feel free to ask.
The administrator has disabled public write access.

Re: Curiosity about voice synthesizer 1 year 9 months ago #832

Ah, so to attempt it with just an Arduino would be quite difficult, if impossible I take it?
The administrator has disabled public write access.

Re: Curiosity about voice synthesizer 1 year 9 months ago #833

I think it would be useful to add some more hardware. There are quite a few audio sample player shields that would help lighten the load. Ideally you would be able to crossfade between two samples to avoid the abrupt phoneme changes and simply control the timing from the Arduino. The sample based approach has its limits as I discussed earlier, which is what make the Babblebot particularly flexible for synthetic voice applications. However, I do encourage you to experiment - you will no doubt learn a lot in your exploration and earn the badge of true Maker.
The administrator has disabled public write access.
The following user(s) said Thank You: bob152637485
  • Page:
  • 1
Time to create page: 0.231 seconds