search

timmahrt / ProMo

Prososdy Morph: A python library for manipulating pitch and duration in an algorithmic way, for resynthesizing speech.
MIT License
81 stars 21 forks source link

Updating ProMo #4

Open sw5813 opened 1 year ago

sw5813 commented 1 year ago

Hey @timmahrt, I see that you had a PR in progress to update ProMo a bit ago, do you know what else needs to be done to push that through? I came across your library and was interested in trying it out, but it looks like it's heavily dependent on PraatIO which has been updated more recently, so the examples in v1.x of ProMo no longer work out of the box. For instance, I was trying to run the examples and hit an error on from praatio import dataio- digging into praatio, it looks like dataio has been renamed to data_points. Thanks!

*Update: it looks like everything still works out of the box as long as I import an older version of praatio, but I'd still be interested in your plans for updating!

timmahrt commented 1 year ago

Oh sorry! I haven't used it recently. I think I know the problem. I can take a look maybe this weekend. Sorry for the headaches!

On Wed, Jul 12, 2023, 04:22 Summer Wu @.***> wrote:

Hey @timmahrt https://github.com/timmahrt, I see that you had a PR in progress to update ProMo a bit ago, do you know what else needs to be done to push that through? I came across your library and was interested in trying it out, but it looks like it's heavily dependent on PraatIO which has been updated more recently, so the examples in v1.x of ProMo no longer work out of the box. For instance, I was trying to run the examples and hit an error on from praatio import dataio- digging into praatio, it looks like dataio has been renamed to data_points. Thanks!

— Reply to this email directly, view it on GitHub https://github.com/timmahrt/ProMo/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAUG7UAHWFFHQ3IH2GZBIGTXPWRXXANCNFSM6AAAAAA2GOPVSE . You are receiving this because you were mentioned.Message ID: @.***>

sw5813 commented 1 year ago

Oh gotcha, thank you! I had one more question- I was looking at your duration_manipulation_example.py and see that TextGrids for the input and output audio files are required. Does ProMo or PraatIO have any tools to get TextGrids, given the audio, or do you have any pointers for how to create TextGrids?

timmahrt commented 1 year ago

Does ProMo or PraatIO have any tools to get TextGrids, given the audio, or do you have any pointers for how to create TextGrids?

Unfortunately ProMo and PraatIO don't have any speech-to-text capabilities. Once you have existing TextGrids, you can use ProMo and PraatIO for manipulating those textgrids (and ProMo for manipulating the audio files).

I've used Julius and SPAAS for doing speech recognition. Kaldi was popular but I never used it. With the boom in AI these days, I guess there are off the shelf systems with robust results but I'm not working in that space anymore unfortunately. 🙈

timmahrt commented 1 year ago

Ok, I've made a new release: promo 2.0. It doesn't include any meaningful changes but it now works with the latest version of praatio and no longer works with python 2.x--so in that sense, it's a breaking change.

If you give it a whirl and have any feedback or anything, please let me know!