Moham
commited on
Commit
•
740a010
1
Parent(s):
bf8f51b
Ar added
Browse files
README.md
CHANGED
|
@@ -2,11 +2,15 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
language:
|
| 4 |
- en
|
|
|
|
| 5 |
tags:
|
| 6 |
-
|
| 7 |
-
|
| 8 |
library_name: metavoice
|
| 9 |
inference: false
|
|
|
|
|
|
|
|
|
|
| 10 |
---
|
| 11 |
|
| 12 |
MetaVoice-1B is a 1.2B parameter base model trained on 100K hours of speech for TTS (text-to-speech). It has been built with the following priorities:
|
|
@@ -40,4 +44,4 @@ We predict EnCodec tokens from text, and speaker information. This is then diffu
|
|
| 40 |
## Optimizations
|
| 41 |
The model supports:
|
| 42 |
1. KV-caching via Flash Decoding
|
| 43 |
-
2. Batching (including texts of different lengths)
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
language:
|
| 4 |
- en
|
| 5 |
+
- ar
|
| 6 |
tags:
|
| 7 |
+
- pretrained
|
| 8 |
+
- text-to-speech
|
| 9 |
library_name: metavoice
|
| 10 |
inference: false
|
| 11 |
+
datasets:
|
| 12 |
+
- HuggingFaceM4/WebSight
|
| 13 |
+
- fka/awesome-chatgpt-prompts
|
| 14 |
---
|
| 15 |
|
| 16 |
MetaVoice-1B is a 1.2B parameter base model trained on 100K hours of speech for TTS (text-to-speech). It has been built with the following priorities:
|
|
|
|
| 44 |
## Optimizations
|
| 45 |
The model supports:
|
| 46 |
1. KV-caching via Flash Decoding
|
| 47 |
+
2. Batching (including texts of different lengths)
|