cinemasetfree.com
xai/grok-text-to-speech
Convert text to natural-sounding speech with xAI's Grok TTS. 5 voices, 20 languages, expressive speech tags, and high-fidelity MP3 / WAV / telephony audio output.
Capabilities
Cost
Community model (estimated from hardware time)
Input Parameters
| Name | Type | Description | Default | Constraints |
|---|---|---|---|---|
text* | string | Text to synthesize into speech. Maximum 15000 characters. Supports inline speech tags like '[pause]', '[laugh]', and wrapping tags like '<whisper>...</whisper>' for expressive delivery. | — | — |
bit_rate | integer | MP3 bit rate in bits per second. Only used when 'output_format' is 'mp3'. Higher bit rates produce better quality at the cost of file size. | 128000 | 320006400096000128000192000 |
language | string | BCP-47 language code for the input text. Set to 'auto' to let the model auto-detect the language. | "auto" | autoenar-EGar-SAar-AEbnzhfrdehiiditjakopt-BRpt-PTrues-MXes-EStrvi |
output_format | string | Audio codec. 'mp3' is best for general use, 'wav' for lossless audio, 'pcm' for raw audio pipelines, 'mulaw'/'alaw' for telephony. | "mp3" | mp3wavpcmmulawalaw |
sample_rate | integer | Audio sample rate in Hz. Higher rates produce better quality at the cost of file size. | 24000 | 80001600022050240004410048000 |
text_normalization | boolean | Normalize written-form text (numbers, abbreviations, symbols) into spoken-form before generating audio. | false | — |
voice | string | Voice to use for synthesis. 'eve' is energetic and upbeat (default), 'ara' is warm and friendly, 'rex' is confident and clear, 'sal' is smooth and balanced, 'leo' is authoritative and strong. | "eve" | eveararexsalleo |
textrequiredstringText to synthesize into speech. Maximum 15000 characters. Supports inline speech tags like '[pause]', '[laugh]', and wrapping tags like '<whisper>...</whisper>' for expressive delivery.
bit_rateintegerMP3 bit rate in bits per second. Only used when 'output_format' is 'mp3'. Higher bit rates produce better quality at the cost of file size.
128000languagestringBCP-47 language code for the input text. Set to 'auto' to let the model auto-detect the language.
"auto"output_formatstringAudio codec. 'mp3' is best for general use, 'wav' for lossless audio, 'pcm' for raw audio pipelines, 'mulaw'/'alaw' for telephony.
"mp3"sample_rateintegerAudio sample rate in Hz. Higher rates produce better quality at the cost of file size.
24000text_normalizationbooleanNormalize written-form text (numbers, abbreviations, symbols) into spoken-form before generating audio.
falsevoicestringVoice to use for synthesis. 'eve' is energetic and upbeat (default), 'ara' is warm and friendly, 'rex' is confident and clear, 'sal' is smooth and balanced, 'leo' is authoritative and strong.
"eve"443e01eb52b1Updated: 7/25/2026115.6K runs