voice_audio_multi/tests/outetts/warm_bench_requests.json

{
  "requests": [
    {
      "name": "tts_cold",
      "text": "This is the first OuteTTS request in a long-lived session.",
      "language": "en",
      "max_tokens": 1024,
      "seed": 1234
    },
    {
      "name": "tts_repeat",
      "text": "This is the first OuteTTS request in a long-lived session.",
      "language": "en",
      "max_tokens": 1024,
      "seed": 1234
    },
    {
      "name": "tts_longform",
      "text": "Long-form synthesis is split by the shared framework text chunker. Each sentence becomes a bounded request while model weights and the cached generation graph stay inside the same long-lived session. The generated audio chunks are appended in order to form one output waveform.",
      "language": "en",
      "seed": 1234,
      "text_chunk_size": 100,
      "text_chunk_mode": "default"
    },
    {
      "name": "clone_cold",
      "text": "This request builds and caches the reference voice profile.",
      "language": "en",
      "voice_ref": "assets/resources/b.wav",
      "reference_text": "Some call me nature. Others call me Mother Nature. I've been here for over 4.5 billion years. 22,500 times longer than you.",
      "reference_language": "en",
      "max_tokens": 1024,
      "seed": 42
    },
    {
      "name": "clone_repeat",
      "text": "This request builds and caches the reference voice profile.",
      "language": "en",
      "voice_ref": "assets/resources/b.wav",
      "reference_text": "Some call me nature. Others call me Mother Nature. I've been here for over 4.5 billion years. 22,500 times longer than you.",
      "reference_language": "en",
      "max_tokens": 1024,
      "seed": 42
    }
  ]
}