falText to speechComing soon

fal-ai/qwen-3-tts/text-to-speech/0.6b

Name: fal-ai/qwen-3-tts/text-to-speech/0.6b
Brand: fal

fal-ai/qwen-3-tts/text-to-speech/0.6b is a text to speech tool from fal that runs in the background and returns audio. Coming soon: listed with its price, not yet open to runs.

This tool is listed but not yet available to run.

Priced by usage

Pricing

$0.091 per 1,000 characters of text. The final charge comes from usage.

Limits

Runs: In the background
Returns: audio
Longest queue wait: 3600s
Run deadline: 1800s
Largest inline input: 1024 KB
Largest input file: 500 MB
Largest output: 4768 MB
Outputs per run: 16

Input

Field	Kind	Default	Description
`text` required	text	—	The text to be converted to speech.
`prompt`	text	—	Optional prompt to guide the style of the generated speech. This prompt will be ignored if a speaker embedding is provided.
`voice`	one of Vivian, Serena, Uncle_Fu, Dylan, Eric, Ryan, Aiden, Ono_Anna, Sohee	—	The voice to be used for speech synthesis, will be ignored if a speaker embedding is provided. Check out the [documentation](https://github.com/QwenLM/Qwen3-TTS/tree/main?tab=readme-ov-file#custom-voice-generate) for each voice's details and which language they primarily support.
`language`	one of Auto, English, Chinese, Spanish, French, German, Italian, Japanese, Korean, Portuguese, Russian	Auto	The language of the voice.
`speaker_voice_embedding_file_url`	text	—	URL to a speaker embedding file in safetensors format, from `fal-ai/qwen-3-tts/clone-voice/0.6b` endpoint. If provided, the TTS model will use the cloned voice for synthesis instead of the predefined voices.
`reference_text`	text	—	Optional reference text that was used when creating the speaker embedding. Providing this can improve synthesis quality when using a cloned voice.
`top_k`	number (whole) 0 to …	50	Top-k sampling parameter.
`top_p`	number 0 to 1	1	Top-p sampling parameter.
`temperature`	number … to 1	0.9	Sampling temperature; higher => more random.
`repetition_penalty`	number 0 to …	1.05	Penalty to reduce repeated tokens/codes.
`subtalker_dosample`	true or false	true	Sampling switch for the sub-talker.
`subtalker_top_k`	number (whole) 0 to …	50	Top-k for sub-talker sampling.
`subtalker_top_p`	number 0 to 1	1	Top-p for sub-talker sampling.
`subtalker_temperature`	number 0 to 1	0.9	Temperature for sub-talker sampling.
`max_new_tokens`	number (whole) 1 to 8192	200	Maximum number of new codec tokens to generate.

Same request in code

curl https://api.spiral.ai/v1/runs \
  -H "Authorization: Bearer $SPIRAL_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "provider": "fal",
    "model": "fal-ai/qwen-3-tts/text-to-speech/0.6b",
    "input": {
      "text": "I feel like I'\''m taking crazy pills! How can something be both a square and a circle at the same time? It defies all logic!"
    }
  }'