Brandon Thomas
f81d40b5bb
Make logging less verbose
2022-09-10 05:42:50 -04:00
Brandon Thomas
ab0180a591
Update s3 bucket client to respect timeout arg
2022-09-09 00:45:44 -04:00
Brandon Thomas
40e92a1b24
Use no-op Redis status logger optionally
2022-09-08 15:22:52 -04:00
Brandon Thomas
bf9bc013df
New dynamic job progress reporter with no-op version
2022-09-08 14:56:35 -04:00
Brandon Thomas
d1e261e633
no longer fail tests (timebomb bump)
2022-09-01 23:43:31 -04:00
Brandon Thomas
4284aeced6
Substantial refactor: pull user system (except mod controls) into crate
...
This has been tested in storyteller-web to ensure it still works
2022-09-01 23:33:21 -04:00
Brandon Thomas
d047fa50a9
Successfully move one "user" endpoint into new package
2022-09-01 17:25:13 -04:00
Brandon Thomas
3477cc395b
refactor
2022-08-29 04:49:51 -04:00
Brandon Thomas
84c616b6f6
New lib and service crates
2022-08-29 00:28:26 -04:00
Brandon Thomas
1da020abd4
Use single connector with list TTS endpoint
2022-08-17 19:41:30 -04:00
Brandon Thomas
68fec1c8da
leaderboard endpoint now uses connection pool (granted, not great)
2022-08-16 21:58:24 -04:00
Brandon Thomas
efe7248ce9
list_tts_model_categories endpoint now uses single connection from pool
2022-08-16 21:29:30 -04:00
Brandon Thomas
6dfe929655
list_tts_categories endpoint now uses single db connection from pool
2022-08-16 21:21:50 -04:00
Brandon Thomas
43f441693e
Update get_tts_models to use single pooled connection
2022-08-16 20:52:29 -04:00
Brandon Thomas
d393ecda34
All queries to user profile controller use a single DB connection
...
- Pull a single connection out of the pool and reuse it
- Update the benchmark code to handle ownership changes (previously,
I think everything was Sync/Send in the closures)
2022-08-16 18:49:25 -04:00
Brandon Thomas
70daacf559
Profile endpoint instrumented
2022-08-16 02:26:00 -04:00
Brandon Thomas
4bd0e09e7d
Move user profile query to query lib
2022-08-16 01:18:47 -04:00
Brandon Thomas
94ed34b5ba
Benchmarking library and begin implementation
2022-08-13 02:27:21 -04:00
Brandon Thomas
4bc33ffea0
tts-inference-job uses custom vocoder (does not cache)
2022-08-10 04:33:38 -04:00
Brandon Thomas
df5e28ec94
query info about vocoders in tts-inference-job
2022-08-10 03:49:15 -04:00
Brandon Thomas
2a10ec5e70
Add custom vocoder info to TTS Model GET endpoint + query
2022-08-10 03:09:16 -04:00
Brandon Thomas
2f360a0475
Update model edit endpoint to include optional vocoder token option
2022-08-10 02:19:49 -04:00
Brandon Thomas
1211262764
vocoder SELECT / get query
2022-08-10 00:57:39 -04:00
Brandon Thomas
419d6bc654
list vocoder endpoint
2022-08-10 00:36:53 -04:00
Brandon Thomas
f81b6091b3
queries / endpoints (WIP)
2022-08-09 22:06:36 -04:00
Brandon Thomas
28512b685e
Vocoder upload pipeline seems to work!
2022-08-09 22:06:36 -04:00
Brandon Thomas
8696613477
generic-download-job now running (need to test)
2022-08-09 22:06:36 -04:00
Brandon Thomas
962510f784
Fix storyteller-web build
2022-08-09 22:06:36 -04:00
Brandon Thomas
47b7c71cdb
WIP generic-download-job
2022-08-09 22:06:36 -04:00
Brandon Thomas
dbd7e949b7
WIP generic-download-job
2022-08-09 22:06:36 -04:00
Brandon Thomas
08c05cbcf2
Schema changes, endpoints, routes, queries, etc.
2022-08-09 22:06:36 -04:00
Brandon Thomas
f8175d7666
reusable_types, WIP endpoints
2022-08-09 22:06:36 -04:00
Brandon Thomas
bf26f7864f
Move last remaining jobs out of storyteller-web and into own crates
...
- w2l-inference-job
- w2l-download-job
These are now in their own independent crates, which should be easier to
do build caching on.
2022-08-04 04:15:55 -04:00
Brandon Thomas
a36394238a
Fix a few broken tests, but this needs revisiting
...
- i18n means this module should be rewritten
- is Spanish even being handled correctly?
2022-08-04 03:10:42 -04:00
Brandon Thomas
76933402b5
Fix tests failing to build
2022-08-04 02:17:20 -04:00
Brandon Thomas
2b9cde68e4
Obscure model text_pipeline_type
2022-07-27 23:12:23 -04:00
Brandon Thomas
6d79f91a0a
improve comments on text pipelines
2022-07-27 23:12:23 -04:00
Brandon Thomas
d458d753fb
Cache W2L templates in-memory
2022-07-27 02:51:48 -04:00
Brandon Thomas
fc88730386
Fix tts job count query
2022-07-25 04:49:36 -04:00
Brandon Thomas
61dd01e5c2
Add endpoint to expose pending TTS jobs publicly to the UI
2022-07-25 02:44:47 -04:00
Brandon Thomas
101d145709
Query pending TTS job count (public use)
2022-07-25 02:19:30 -04:00
Brandon Thomas
820417ac0f
More CORS domains
2022-07-25 01:57:41 -04:00
Brandon Thomas
b9fed6a933
Cache model lookups, removing some query load from DB
2022-07-14 20:03:32 -04:00
Brandon Thomas
3ae65f5d11
Add multi-item cache
2022-07-14 19:36:23 -04:00
Brandon Thomas
b807adc73d
Create "memory_caching" lib crate, start building more tools there
2022-07-14 19:03:22 -04:00
Brandon Thomas
b71278aef1
Push rollout into the future to fix the pronunciation bug
...
We rolled this out as the default too soon. New models are trained on
10% arpabet only.
2022-07-06 18:55:15 -04:00
Brandon Thomas
b801f7c19e
Edit TTS model with text pipeline type
2022-07-06 01:51:08 -04:00
Brandon Thomas
f75f55fa22
Use Rust Enum instead
2022-07-06 01:28:08 -04:00
Brandon Thomas
152a2bfad7
Move queries out of edit TTS model handler.
2022-07-06 00:08:36 -04:00
Brandon Thomas
03c6f1a269
Use HTTP inference arg for text_pipeline_type
2022-07-05 21:07:37 -04:00