Update README.md
Browse files
README.md
CHANGED
@@ -53,9 +53,10 @@ It models audio as tokens and can generate high-quality audio with consistent st
|
|
53 |
### Key features
|
54 |
|
55 |
1. Extremely small, based on GPT-2 small architecture. The methodology can be extended to any autoregressive transformer-based architecture.
|
56 |
-
2.
|
57 |
-
|
58 |
-
|
|
|
59 |
|
60 |
### Details
|
61 |
|
@@ -64,10 +65,6 @@ It models audio as tokens and can generate high-quality audio with consistent st
|
|
64 |
3. Language Support: English, Hindi
|
65 |
4. License: CC BY 4.0
|
66 |
|
67 |
-
### Speed
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
## Technical details
|
72 |
|
73 |
Here's a brief of how the model works:
|
|
|
53 |
### Key features
|
54 |
|
55 |
1. Extremely small, based on GPT-2 small architecture. The methodology can be extended to any autoregressive transformer-based architecture.
|
56 |
+
2. Ultra-fast. Using our [self hosted service option](#self-hosted-service), the model can achieve speeds up to 400 toks/s (4s of audio generation per s) and under 20ms time to first token on RTX6000Ada NVIDIA GPU.
|
57 |
+
1. On RTX6000Ada, it can support a batch size of 1k with full context length of 1024 tokens
|
58 |
+
3. Supports voice cloning with small prompts (<5s).
|
59 |
+
4. Code mixing text input in 2 languages - English and Hindi.
|
60 |
|
61 |
### Details
|
62 |
|
|
|
65 |
3. Language Support: English, Hindi
|
66 |
4. License: CC BY 4.0
|
67 |
|
|
|
|
|
|
|
|
|
68 |
## Technical details
|
69 |
|
70 |
Here's a brief of how the model works:
|