docs: binary now prints text (in-binary detok)
Browse files
README.md
CHANGED
|
@@ -27,9 +27,10 @@ GGUF build of **[SenseVoiceSmall](https://huggingface.co/FunAudioLLM/SenseVoiceS
|
|
| 27 |
| `sensevoice-small.gguf` | 936 MB | f32 reference |
|
| 28 |
|
| 29 |
## Usage
|
|
|
|
| 30 |
```bash
|
| 31 |
# 1. get the VAD too (for long audio): huggingface-cli download FunAudioLLM/fsmn-vad-GGUF
|
| 32 |
-
llama-funasr-sensevoice -m sensevoice-small-f16.gguf -a audio.wav --vad fsmn-vad.gguf
|
| 33 |
```
|
| 34 |
On CPU (8 threads) this reaches **8.01 % CER** on the 184-clip Mandarin benchmark — vs whisper.cpp 22–31 %. See the [benchmark](https://github.com/FunAudioLLM/SenseVoice/blob/main/runtime/llama.cpp/BENCHMARKS.md).
|
| 35 |
|
|
|
|
| 27 |
| `sensevoice-small.gguf` | 936 MB | f32 reference |
|
| 28 |
|
| 29 |
## Usage
|
| 30 |
+
The binary prints **transcription text** directly (no Python detok). `--ids` for raw ids / `--keep-tags` for the lang/emotion tags.
|
| 31 |
```bash
|
| 32 |
# 1. get the VAD too (for long audio): huggingface-cli download FunAudioLLM/fsmn-vad-GGUF
|
| 33 |
+
llama-funasr-sensevoice -m sensevoice-small-f16.gguf -a audio.wav --vad fsmn-vad.gguf
|
| 34 |
```
|
| 35 |
On CPU (8 threads) this reaches **8.01 % CER** on the 184-clip Mandarin benchmark — vs whisper.cpp 22–31 %. See the [benchmark](https://github.com/FunAudioLLM/SenseVoice/blob/main/runtime/llama.cpp/BENCHMARKS.md).
|
| 36 |
|