ekoc
446c2f6814
Updating to CUDA 12.8, pytorch 2.7 and deepseepd 0.17.2
2025-07-11 20:33:20 +02:00
erew123
35dcd9710c
Update F5 to use new models & Inference
...
Adds the new F5 models
Updates inference code
Sets speed to 1 (when not specified)
2025-04-09 09:23:38 +01:00
ekoc
5c017f8752
Provide config setting for RVC model cache size
2025-03-19 21:23:56 +01:00
S-T-K
f8516e5a14
maxsize 3 for LRU cache for RVC
...
Feature was introduced in https://github.com/erew123/alltalk_tts/pull/383 but without maxsize, which meant rvc models would never get unloaded
2025-03-18 09:33:39 +01:00
ekoc
8a5db3a687
Improving docker build time by downloading the TTS model during build time, thereby enhancing firstrun.py
2025-03-09 16:07:05 +01:00
Андрей Петров
90af3c2ec5
prevent variable "model_name" from UnboundLocalError
2025-03-03 19:11:09 +03:00
erew123
698efdcdec
Update RVC Hubert Loader - add_safe_globals fix
2025-02-24 15:06:05 +00:00
ekoc
284c100860
Refactoring docker images, updating CUDA to 12.6 and pytorch 2.6
2025-02-12 22:01:34 +01:00
erew123
591d86978b
Update rvc_files.json
2025-01-30 06:26:44 +00:00
Ilyas Hilali
0182e8fd01
Add a warning when a pickle file is loaded
2025-01-05 12:13:07 +01:00
Ilyas Hilali
e72a7bf5c9
Add initial support for pickletensor models to F5-TTS
...
* Tested with @RASPAUDIO french model available here : https://huggingface.co/RASPIAUDIO/F5-French-MixedSpeakers-reduced
2025-01-03 18:32:52 +01:00
Azalea Gui
0c6e113aff
[F] Fix #251 : Expected a value of type 'str'
2024-11-28 22:19:44 -05:00
erew123
231bc21fa7
Updated to use new ffmpeg
2024-11-25 23:33:12 +00:00
erew123
7909327f51
fix missing model logic
2024-11-24 06:18:24 +00:00
erew123
03c9d810ff
fix missing model logic
2024-11-24 06:17:53 +00:00
erew123
e0621a97d6
Added JSON latents creation
...
- System automatically creates `.json` latent files alongside voice samples
- Latents are voice characteristics extracted from audio
- Generated on first use of a voice file
- Stored next to original audio (e.g., `broadcaster_male.wav` → `broadcaster_male.json`)
- Improves generation speed for subsequent uses
- No manual management needed
2024-11-24 05:56:54 +00:00
erew123
925c73af63
Update help_content.py
2024-11-22 08:14:58 +00:00
erew123
85f204c3f1
Updates to Piper
2024-11-22 07:33:28 +00:00
erew123
1a5c10710d
Updates to TTS Engines
2024-11-22 07:32:38 +00:00
erew123
4aa4493cc5
Clearer messages commit correction
2024-11-11 11:35:59 +00:00
erew123
5a3a20dcba
Clearer message on missing model files
2024-11-11 11:33:40 +00:00
erew123
590eb23e78
add UNetT import
2024-11-05 10:25:32 +00:00
erew123
1b4e982955
add back setup routine
2024-11-05 10:22:50 +00:00
erew123
5a81e4818b
Add support for the E2 model to F5-TTS
2024-11-05 10:19:45 +00:00
erew123
1a5945235d
F5-TTS Unload model correct tab
2024-11-05 09:51:50 +00:00
erew123
0e61d0aa24
Added F5-TTS engine
2024-11-05 00:15:04 +00:00
erew123
834f7a4454
Added F5-TTS engine
2024-11-05 00:12:44 +00:00
erew123
2c2c31ef09
F5-TTS final updates
2024-11-05 00:03:47 +00:00
erew123
5d43b23eb7
F5-TTS Engine/Settings changes
2024-11-04 22:27:19 +00:00
erew123
bcd84a4c20
Setting for float16
...
# Set Parler Precision. Options are:
# "float32" (full precision)
# "float16" (half precision)
# "bfloat16" (brain floating point)
2024-11-03 20:24:24 +00:00
erew123
a56a2bae5c
f5tts - still a work in progress
...
Just uploading this so I have a reference point in time that I wont lose.
2024-11-03 15:11:50 +00:00
ekoc
362683a645
Introduce LRU cache for RVC
2024-10-18 09:21:15 +02:00
erew123
23d813021b
Update xtts_settings_page.py
2024-10-16 07:57:30 +01:00
erew123
5ced67d5ee
Update vits_settings_page.py
2024-10-16 07:57:10 +01:00
erew123
b19b6f9f48
Update piper_settings_page.py
2024-10-16 07:56:52 +01:00
erew123
97ddb39218
Update parler_settings_page.py
2024-10-16 07:56:32 +01:00
erew123
76a50cf47f
Note re Hindi support on XTTS
...
Hindi support only works on XTTS model 2.0.3 and must use the API method (load it as APITTS) in the Generate window of the Gradio interface.
2024-10-01 14:50:52 +01:00
erew123
b7aa3a7c5c
Multi audio sample TTS generation w/XTTS
...
🟧 Using Single Voice Samples
Voice samples are stored in `/alltalk_tts/voices/` and should be named using the following format `name.wav`. These files will be listed as `name.wav` in the available voices list.
🟧 Using Multiple Voice Samples
If you have multiple voice samples for a single voice, you can organize them into subfolders within the `/alltalk_tts/voices/` directory. Each subfolder should be named according to the voice it contains, up to 5 voice samples will be randomly selected for use.
• Each subfolder should reflect the name or type of the voice it contains (e.g., female_voice, male_voice).
• The voice samples inside each subfolder should follow the standard .wav format.
• An example folder path would be `/alltalk_tts/voices/mynewvoice/` and this would be listed in the available voices list as `mynewvoice/`.
This organization allows for easy selection and management of multiple voice samples while ensuring the system can correctly identify and utilize each voice. Manual CURL API requests would send the folder in the format `mynewvoice/`.
2024-08-30 12:55:06 +01:00
erew123
c1c7b8cf65
Updated Parler to the v1 model (mini and large available)
2024-08-09 10:32:47 +01:00
erew123
df764f71ef
update Parler model to v1 mini
2024-08-09 08:46:05 +01:00
bmalaski
a97ad658c1
ported changes form original branch
2024-06-20 22:54:57 -04:00
erew123
bc0ebc03a6
Removed ryan high and medium
...
Both those models have issues with generation https://github.com/rhasspy/piper/issues/128
2024-06-11 10:18:17 +01:00
erew123
493a603542
Update tts_engines.json
2024-06-06 22:51:03 +01:00
erew123
8538b96cd2
Add files via upload
2024-06-06 22:41:01 +01:00
erew123
9fd928f6e7
initial upload attempt
...
initial upload attempt
2024-06-06 21:40:12 +01:00
ekoc