feat: add phoneme synthesis capabilities and enhance TTS functionality#53
feat: add phoneme synthesis capabilities and enhance TTS functionality#53stretchyboy wants to merge 2 commits into
Conversation
stretchyboy
commented
Jun 15, 2026
- Introduced new phoneme synthesis methods in Dj class: speak_phoneme and speak_phoneme_chunk for generating audio samples from ARPAbet phonemes.
- Implemented phoneme chunking logic to maintain word boundaries during synthesis.
- Added a new phonemize module for ARPAbet to IPA conversion and phoneme chunking.
- Updated web routes to handle phoneme generation requests and caching of responses.
- Enhanced HTML templates with versioning for asset management.
- Added tests for sentence chunk planning to ensure correct isolation of words during synthesis.
- Included pronouncing library for stress pattern extraction from ARPAbet strings.
- Introduced new phoneme synthesis methods in Dj class: speak_phoneme and speak_phoneme_chunk for generating audio samples from ARPAbet phonemes. - Implemented phoneme chunking logic to maintain word boundaries during synthesis. - Added a new phonemize module for ARPAbet to IPA conversion and phoneme chunking. - Updated web routes to handle phoneme generation requests and caching of responses. - Enhanced HTML templates with versioning for asset management. - Added tests for sentence chunk planning to ensure correct isolation of words during synthesis. - Included pronouncing library for stress pattern extraction from ARPAbet strings.
|
OK that's really cool @stretchyboy ! Thanks for this improvement suggestion, it speaks to me :) |
|
Hi,
I wasn't actually intending in doing a PR straight away but was going to try performing with it first but the button popped up and I started it.
Yes there is AI assisted code in there. There isn't on the Strudel side but I was having trouble with some bits of it and got some help. I broke my hip and am in painkillers so I would check it carefully.
I do only have it running locally it will be tomorrow morning before I can do anything else with it.
Tar.
Sent from Outlook for Android<https://aka.ms/AAb9ysg>
…________________________________
From: Alexandre G.-Raymond ***@***.***>
Sent: Monday, 15 June 2026 20:40:41
To: ilesinge/shabda ***@***.***>
Cc: Martyn Eggleton ***@***.***>; Mention ***@***.***>
Subject: Re: [ilesinge/shabda] feat: add phoneme synthesis capabilities and enhance TTS functionality (PR #53)
[https://avatars.githubusercontent.com/u/501674?s=20&v=4]ilesinge left a comment (ilesinge/shabda#53)<#53 (comment)>
OK that's really cool @stretchyboy<https://github.com/stretchyboy> ! Thanks for this improvement suggestion, it speaks to me :)
For disclaimer purposes, did you use AI? (not that I reject it but I would review the PR more carefully)
Did you deploy any test version somewhere public (that I could test) or is it just local? Could you share a screenshot?
I'll test it locally though if no public test version is deployed.
Thanks again!
—
Reply to this email directly, view it on GitHub<#53?email_source=notifications&email_token=AABX7TV5LBMPCZDHMC4A23D5ABGLTA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTINZRGE3TIMBVGQ32M4TFMFZW63VHNVSW45DJN5XKKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-4711740547>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AABX7TSKKSHWAH2E2QNNOED5ABGLTAVCNFSNUABFKJSXA33TNF2G64TZHMZDAMRSGEYDGMZSHNEXG43VMU5TINRWGU2DIMZSGE4KC5QC>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
Sorry to hear about your hip, get well soon! |
|
I don't seem to have understood how the branches work with pull requests and it's accidentally sucked in a separate change on a totally different project. https://codeberg.org/uzu/strudel/pulls/2059/commits/aac01e4293da74d490404e739ff8048adc142a59. How do I take this back out? |
I think I have now fixed that. |
|
I'm putting it up on codespaces for more testing and hit a few issues. |
|
I can't work out how to do the TTS api authorization on codespaces but the sentence chunking is working here |