fal
fal.ai open-source TTS models through the Speechbase gateway, with voice cloning on f5-tts.
| Prefix | fal-ai |
| Default model | (specify a model) |
| Provider key | Connect under Provider Keys |
fal hosts open-source TTS models, so there's no default — you must name a model.
Route to it by prefixing the model with fal-ai/.
Models
| Model | Streaming | Voice cloning | Open source | Languages |
|---|---|---|---|---|
f5-tts | — | Yes | Yes | 9 |
kokoro | — | — | Yes | 5 |
orpheus-tts | — | — | Yes | 7 |
Streaming isn't available — the gateway returns the complete audio once it's generated.
The model string
The gateway model string is fal-ai/<model> — a single fal-ai prefix,
for example fal-ai/f5-tts. The gateway parses exactly one slash, so a doubled
form like fal-ai/fal-ai/f5-tts is rejected.
Usage
curl -X POST https://api.speechbase.ai/v1/audio/speech \
-H "Authorization: Bearer $SPEECHBASE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"mode": "inline",
"model": "fal-ai/kokoro",
"voice": "default",
"text": "Hello from Speechbase!",
"output": "mp3"
}' --output hello.mp3Voice cloning
f5-tts clones from a reference sample. Because inline requests take a plain
voice string only, register the reference as a saved
Voice and address it by voiceId with
mode: "voice".
Provider options
Anything in providerOptions is forwarded to the fal model endpoint unchanged.