[NFC] Update misleading seed-tts-eval optional dependency guidance (#4619)
Signed-off-by: Yuanheng <jonathan.zhaoyh@gmail.com>
This commit is contained in:
1 parent
50bc51ffdf
commit
9177b37e2f
3 files changed
+6
-6
No files matched your search
@@ -119,7 +119,7 @@ def add_seed_tts_cli_args(parser: argparse.ArgumentParser) -> None:
|
||||
help="Keep synthesized audio as 24 kHz mono PCM for WER (works with "
|
||||
"--backend openai-audio-speech or openai-chat-omni). Scoring follows "
|
||||
"zhaochenyang20/seed-tts-eval (Whisper-large-v3 / Paraformer-zh + jiwer). "
|
||||
"Sets SEED_TTS_WER_EVAL=1. Install: pip install 'vllm-omni[seed-tts-eval]'. "
|
||||
"Sets SEED_TTS_WER_EVAL=1. Install: pip install 'vllm-omni[dev]'. "
|
||||
"Optional: SEED_TTS_EVAL_DEVICE, SEED_TTS_HF_WHISPER_MODEL.",
|
||||
)
|
||||
g.add_argument(
|
||||
|
||||
Reference in new issue
Block a user