Models

Reson8 provides multiple speech-to-text models to choose from. Models make different trade-offs between transcription quality, price, and speed, so you can pick the one that fits your use case.

Every request is transcribed by one model. Pass the model query parameter to choose it; when omitted, the request uses resonant-1.

?model=rumble-2-flash

Available Models

Model	Slug	Prerecorded	Realtime	Turns
Rumble-2-Flash	rumble-2-flash	Supported	-	-
Resonant-1	resonant-1	Supported	Supported	Supported

Rumble-2-Flash

A model optimized for speed on the Prerecorded API: upload a file and get its plain transcript back. It is a good fit when you need just the text of a recording, without speaker labels, timestamps, or vocabulary customisation.

Languages: Afrikaans, Albanian, Arabic, Basque, Bengali, Bosnian, Cantonese, Catalan, Cebuano, Chinese, Croatian, Czech, Danish, Dutch, English, Filipino, Finnish, French, German, Greek, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Korean, Latvian, Lithuanian, Macedonian, Malay, Maltese, Norwegian Bokmål, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Spanish, Swedish, Thai, Turkish, Ukrainian, and Vietnamese. See Languages for the language codes.

Supported parameters

Supports encoding, sample_rate, channels, and language. Requests with diarize, include_timestamps, include_words, phrases, custom_model_id, patterns, or filler_mode are rejected, and a custom model configured on the API client is not applied.

Resonant-1

The default model, available on every endpoint and with every feature: diarization, timestamps and word-level detail, custom models and phrases, patterns, and filler mode. Use it unless you have a reason to pick another model.

Languages: Dutch, English, French, Frisian, German, Italian, Polish, Portuguese, Spanish, and Swedish. See Languages for the language codes.