{
"input": [
{
"id": "ffd452c5",
"content": [
{
"audio": "[BASE64_AUDIO: wav, ~183.0KB]",
"format": "wav"
},
{
"text": "Use the reference audio and prompt transcript to synthesize the target text in the same speaker voice.\nPrompt transcript: We asked over twenty different people, and they all said it was his.\nTarget text: Get the trust fund to the bank early.\nReturn only the generated audio."
}
]
}
],
"target": "Get the trust fund to the bank early.",
"id": 0,
"group_id": 0,
"subset_key": "en",
"metadata": {
"filename": "common_voice_en_10119832-common_voice_en_10119840",
"prompt_text": "We asked over twenty different people, and they all said it was his.",
"text": "Get the trust fund to the bank early.",
"prompt_audio_path": "prompt-wavs/common_voice_en_10119832.wav",
"reference_audio_path": "wavs/common_voice_en_10119832-common_voice_en_10119840.wav",
"language": "en",
"wer_language": "en"
}
}
提示模板
提示模板:
Use the reference audio and prompt transcript to synthesize the target text in the same speaker voice.
Prompt transcript: {prompt_text}
Target text: {text}
Return only the generated audio.
使用方法
使用 CLI
evalscope eval\
--model YOUR_MODEL \
--api-url OPENAI_API_COMPAT_URL \
--api-key EMPTY_TOKEN \
--datasets seed_tts_eval \
--limit 10 # Remove this line for formal evaluation
使用 Python
fromevalscopeimportrun_taskfromevalscope.configimportTaskConfigtask_cfg = TaskConfig(
model='YOUR_MODEL',
api_url='OPENAI_API_COMPAT_URL',
api_key='EMPTY_TOKEN',
datasets=['seed_tts_eval'],
dataset_args={
'seed_tts_eval': {
# subset_list: ['en', 'zh'] # optional, evaluate specific subsets
}
},
limit=10, # Remove this line for formal evaluation
)
run_task(task_cfg=task_cfg)