update README.md for v1.0.0
Browse files- README.md +8 -5
- src/f5_tts/infer/README.md +1 -1
- src/f5_tts/train/README.md +2 -2
README.md
CHANGED
@@ -159,9 +159,8 @@ volumes:
|
|
159 |
```bash
|
160 |
# Run with flags
|
161 |
# Leave --ref_text "" will have ASR model transcribe (extra GPU memory usage)
|
162 |
-
f5-tts_infer-cli \
|
163 |
-
--
|
164 |
-
--ref_audio "ref_audio.wav" \
|
165 |
--ref_text "The content, subtitle or transcription of reference audio." \
|
166 |
--gen_text "Some text you want TTS model generate for you."
|
167 |
|
@@ -182,15 +181,19 @@ f5-tts_infer-cli -c src/f5_tts/infer/examples/multi/story.toml
|
|
182 |
|
183 |
## Training
|
184 |
|
185 |
-
### 1.
|
186 |
|
187 |
-
|
|
|
|
|
188 |
|
189 |
```bash
|
190 |
# Quick start with Gradio web interface
|
191 |
f5-tts_finetune-gradio
|
192 |
```
|
193 |
|
|
|
|
|
194 |
|
195 |
## [Evaluation](src/f5_tts/eval)
|
196 |
|
|
|
159 |
```bash
|
160 |
# Run with flags
|
161 |
# Leave --ref_text "" will have ASR model transcribe (extra GPU memory usage)
|
162 |
+
f5-tts_infer-cli --model F5TTS_v1_Base \
|
163 |
+
--ref_audio "provide_prompt_wav_path_here.wav" \
|
|
|
164 |
--ref_text "The content, subtitle or transcription of reference audio." \
|
165 |
--gen_text "Some text you want TTS model generate for you."
|
166 |
|
|
|
181 |
|
182 |
## Training
|
183 |
|
184 |
+
### 1. With Hugging Face Accelerate
|
185 |
|
186 |
+
Refer to [training & finetuning guidance](src/f5_tts/train) for best practice.
|
187 |
+
|
188 |
+
### 2. With Gradio App
|
189 |
|
190 |
```bash
|
191 |
# Quick start with Gradio web interface
|
192 |
f5-tts_finetune-gradio
|
193 |
```
|
194 |
|
195 |
+
Read [training & finetuning guidance](src/f5_tts/train) for more instructions.
|
196 |
+
|
197 |
|
198 |
## [Evaluation](src/f5_tts/eval)
|
199 |
|
src/f5_tts/infer/README.md
CHANGED
@@ -77,7 +77,7 @@ f5-tts_infer-cli \
|
|
77 |
f5-tts_infer-cli --model F5TTS_Base --vocoder_name bigvgan --load_vocoder_from_local
|
78 |
|
79 |
# Use custom path checkpoint, e.g.
|
80 |
-
f5-tts_infer-cli --ckpt_file ckpts/
|
81 |
|
82 |
# More instructions
|
83 |
f5-tts_infer-cli --help
|
|
|
77 |
f5-tts_infer-cli --model F5TTS_Base --vocoder_name bigvgan --load_vocoder_from_local
|
78 |
|
79 |
# Use custom path checkpoint, e.g.
|
80 |
+
f5-tts_infer-cli --ckpt_file ckpts/F5TTS_v1_Base/model_1250000.safetensors
|
81 |
|
82 |
# More instructions
|
83 |
f5-tts_infer-cli --help
|
src/f5_tts/train/README.md
CHANGED
@@ -40,10 +40,10 @@ Once your datasets are prepared, you can start the training process.
|
|
40 |
accelerate config
|
41 |
|
42 |
# .yaml files are under src/f5_tts/configs directory
|
43 |
-
accelerate launch src/f5_tts/train/train.py --config-name
|
44 |
|
45 |
# possible to overwrite accelerate and hydra config
|
46 |
-
accelerate launch --mixed_precision=fp16 src/f5_tts/train/train.py --config-name
|
47 |
```
|
48 |
|
49 |
### 2. Finetuning practice
|
|
|
40 |
accelerate config
|
41 |
|
42 |
# .yaml files are under src/f5_tts/configs directory
|
43 |
+
accelerate launch src/f5_tts/train/train.py --config-name F5TTS_v1_Base.yaml
|
44 |
|
45 |
# possible to overwrite accelerate and hydra config
|
46 |
+
accelerate launch --mixed_precision=fp16 src/f5_tts/train/train.py --config-name F5TTS_v1_Base.yaml ++datasets.batch_size_per_gpu=19200
|
47 |
```
|
48 |
|
49 |
### 2. Finetuning practice
|