Developing ad hoc stt and tts systems

[JRMeyer]

[
[[Vincent_BERMENT
[December 17, 2017, 9:02pm]

Dear all,

I would like to develop a dedicated STT system. More precisely, I want slash
to make a system that recognizes (and also generates) a 'small' number slash
of sentences ( slash ~5,000) but perfectly. Would it be possible with
DeepSpeech ?

I was also wondering whether such systems would require a GPU server tu slash
work or if they could run on small platforms such as Paspberry Pi 3 B.

Thank you for your help !

Vincent

[This is an archived TTS discussion thread from discourse.mozilla.org/t/developing-ad-hoc-stt-and-tts-systems]

[JRMeyer]

>>> elpimous_robot
[December 17, 2017, 10:51pm]

Hi, Vincent,

First, about the RPI3 usage, inferences with large model doesn't
seems to be possible, at least for now because inference process takes
too long time, to hope realtime. slash
Lissyx is working on AOT model (optimized), so perhaps it could be
possible, reducing the model !

You'd like to use a model, with a 5000 possible sentences, with the
best accuracy possible : slash
Sure, with Deepspeech, you're in the right place.

Hoping that your sentences contain standard words, slash
you create a vocab.txt, containing all your slash ~5000 sentences, a LM and a
TRIE files. slash
You should obtain a very good accuracy, quite better than everything
known in phonemes STT slash
But, I must say that, actually, there is no perfect solution ! Not
yet...

Sorry, Deepspeech is only, actually a STT engine.

Hope it helped

[Archived Post]

[JRMeyer]

>>> erogol
[January 5, 2018, 3:45pm]

What I can adhere is, we also begin our research on TTS.

[Archived Post]

[JRMeyer]

>>> yv001
[March 8, 2018, 8:25am]

At what stage is the TTS
project? are there any online resources, repos, forums etc. about the
mozillian TTS?

[Archived Post]

[JRMeyer]

>>> jtane
[March 8, 2018, 4:01pm]

One thing I have to wonder about this question is:

model size

I wonder how difficult it would be to shardize language models. Actually
that sounds like a promissing approach.

[Archived Post]

[JRMeyer]

>>> elpimous_robot
[March 8, 2018, 8:54pm]

slash
Well, quality model is dependant of learnt material volume. slash
Sure it must be good quality material. slash
the more material (wav sentences/corresponding transcritions), the
better the model inferences ! slash
Of course, the more material == the bigger the model !

[Archived Post]

[JRMeyer]

>>> erogol
[June 21, 2019, 8:19am]

{.site-icon width='16'

{.thumbnail

### mozilla/TTS

Deep learning for Text to Speech . Contribute to mozilla/TTS development
by creating an account on GitHub.

slash
https://discourse.mozilla.org/c/tts

[Archived Post]