readme/notes

.gitignore

@@ -1,3 +1,4 @@
+*.pyc
 *.swo
 *.swp
 *secret*sh
@@ -13,3 +14,10 @@ production_database_secret_query.sh
 secrets.toml
 secrets.yaml
 target/
+
+# Models for testing
+future-todo/ControllableTalkNet/models
+tts/hifigan/hifisr
+
+# Other garbage to prune/remove
+tts/hifigan/LJSpeech-1.1/

README.md

@@ -1,6 +1,40 @@
 tts-core
 ========
-
 The core TTS models that power FakeYou.
 
 I'm in the progress of pulling these out of the service monorepo so we can iterate faster.
+
+TODO
+----
+
+Core cleanup
+
+- [ ] Cleanup Tacotron code 
+  - [ ] Integrate WaveGlow and HifiGan more cleanly
+  - [ ] Define the HTTP server interface
+  - [ ] Define the model checking interface
+  - [ ] Clean up in-memory LRU caching
+  - [ ] There are two copies of Nemo. Why? Clean this up.
+- [ ] Deployability
+  - [ ] Clean up requirements.txt and upgrade packages 
+  - [ ] Clean up Dockerfile
+
+Make the models sound better
+
+- [ ] Arpabet
+  - [ ] G2P
+  - [ ] Arpabet custom dictionary
+  - [ ] User-supplied Arpabet strings in braces
+
+- [ ] Word break down 888 -> eight hundred eighty eight
+
+- [ ] Support I18N
+  - [ ] Accent character support
+
+Introduce new models
+
+- [ ] TalkNet
+- [ ] GlowTTS + MelGan (vocoder)
+- [ ] Make these work under the same venv
+
+Lots more I haven't covered yet...

future-todo/README.md

@@ -0,0 +1,9 @@
+TalkNet
+=======
+TalkNet is not currently in a usable state. Ideally we can have both Tacotron and 
+Talknet models operate at the same time from the same uniform codebase.
+
+The reason we want them to work together is so that we can leverage the same GPU
+resources and use LRU in-memory caching.
+
+This copy of TalkNet was taken from a public MIT-licensed repo and notebook (TODO: link)