Spaces:
Running
on
Zero
Running
on
Zero
Update README.md
Browse files
README.md
CHANGED
@@ -1,14 +1,16 @@
|
|
1 |
---
|
2 |
title: Kokoro TTS Zero
|
3 |
emoji: 📊
|
4 |
-
colorFrom:
|
5 |
-
colorTo:
|
6 |
sdk: gradio
|
7 |
sdk_version: 5.9.1
|
8 |
app_file: app.py
|
9 |
-
pinned:
|
10 |
license: apache-2.0
|
11 |
short_description: A100 GPU Accelerated Inference applied to Kokoro-82M TTS
|
|
|
|
|
12 |
---
|
13 |
|
14 |
# Kokoro TTS Demo Space
|
@@ -24,16 +26,10 @@ This Space provides a Gradio interface for the Kokoro TTS model, allowing users
|
|
24 |
|
25 |
## Technical Details
|
26 |
|
27 |
-
-
|
28 |
- Dynamically loads required modules from hexgrad/Kokoro-82M repository
|
29 |
-
- Automatically downloads model and voice files from Hugging Face Hub
|
30 |
-
- Implements proper GPU memory handling
|
31 |
-
- Includes caching in /data/.huggingface for faster restarts
|
32 |
|
33 |
## Dependencies
|
34 |
-
|
35 |
-
The Space uses modules from two repositories:
|
36 |
-
- remsky/Kokoro-FastAPI: This repository (UI and Zero GPU implementation)
|
37 |
- hexgrad/Kokoro-82M: Original model repository (core TTS functionality)
|
38 |
|
39 |
All dependencies are automatically handled:
|
@@ -46,27 +42,22 @@ All dependencies are automatically handled:
|
|
46 |
- Python 3.10.13
|
47 |
- PyTorch 2.2.2
|
48 |
- Gradio 5.9.1
|
49 |
-
- Zero GPU
|
50 |
|
51 |
## Available Voices
|
52 |
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
- bf_isabella: Young Professional, British
|
61 |
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
- bm_lewis: Assured and Raspy, British
|
67 |
|
68 |
## Notes
|
69 |
-
|
70 |
-
- First generation may take longer due to model initialization
|
71 |
-
- GPU is allocated only during speech generation
|
72 |
-
- Model and voices are cached in /data/.huggingface for faster subsequent runs
|
|
|
1 |
---
|
2 |
title: Kokoro TTS Zero
|
3 |
emoji: 📊
|
4 |
+
colorFrom: gray
|
5 |
+
colorTo: purple
|
6 |
sdk: gradio
|
7 |
sdk_version: 5.9.1
|
8 |
app_file: app.py
|
9 |
+
pinned: true
|
10 |
license: apache-2.0
|
11 |
short_description: A100 GPU Accelerated Inference applied to Kokoro-82M TTS
|
12 |
+
models:
|
13 |
+
- hexgrad/Kokoro-82M
|
14 |
---
|
15 |
|
16 |
# Kokoro TTS Demo Space
|
|
|
26 |
|
27 |
## Technical Details
|
28 |
|
29 |
+
- Zero GPU for efficient GPU resource management
|
30 |
- Dynamically loads required modules from hexgrad/Kokoro-82M repository
|
|
|
|
|
|
|
31 |
|
32 |
## Dependencies
|
|
|
|
|
|
|
33 |
- hexgrad/Kokoro-82M: Original model repository (core TTS functionality)
|
34 |
|
35 |
All dependencies are automatically handled:
|
|
|
42 |
- Python 3.10.13
|
43 |
- PyTorch 2.2.2
|
44 |
- Gradio 5.9.1
|
45 |
+
- A100 Zero GPU Enabled
|
46 |
|
47 |
## Available Voices
|
48 |
|
49 |
+
af: Default
|
50 |
+
af_sky: Classic
|
51 |
+
af_bella: Warm
|
52 |
+
af_nicole: Soothing
|
53 |
+
af_sarah: Polished
|
54 |
+
bf_emma: Contemplative
|
55 |
+
bf_isabella: Poised
|
|
|
56 |
|
57 |
+
am_adam: Resonant
|
58 |
+
am_michael: Sincere
|
59 |
+
bm_george: Distinguished
|
60 |
+
bm_lewis: Gravelly
|
|
|
61 |
|
62 |
## Notes
|
63 |
+
- Model Warm-Up takes some time, it shines at longer lengths.
|
|
|
|
|
|