W
Minimax Speech 2.8 HD
Minimax Speech 2.8 HD, a text-to-speech model from WaveSpeed.
speechModelOverride: { model: "minimax-speech-2.8-hd" }
MiniMax·served via
WaveSpeed·Released January 2026Pricing
- List price
- $100 / 1M characters
- Per generation
- $0.00010
Specifications
- Context window
- 10,000 tokens
- Released
- January 2026
- Model string
- minimax/speech-2.8-hd
Parameters
Model options, passed in
speechModelOverride.config.| Parameter | Type | Default | Range / options |
|---|---|---|---|
voice_idVoice preset to use for speech synthesis. | Select | Friendly_Person | Wise_WomanFriendly_PersonInspirational_girlDeep_Voice_ManCalm_WomanCasual_Guy+11 |
speedSpeech speed multiplier. | Number | 1 | 0.5–2, step 0.1 |
volumeVolume level. | Number | 1 | 0.1–2, step 0.1 |
pitchPitch adjustment. | Number | 0 | -12–12, step 1 |
emotionEmotional tone of the speech delivery. | Select | — | happysadangryfearfuldisgustedsurprised+1 |
sample_rateAudio sample rate in Hz. | Select | 44100 | 16000240003200044100 |
bitrateAudio bitrate in bits per second. | Select | 128000 | 3200064000128000256000 |
channelAudio channel configuration. | Select | — | 12 |
formatOutput audio format. | Select | — | mp3wavflac |
language_boostBoost recognition for a specific language. | Select | — | autoAfrikaansArabicBulgarianCatalanChinese+35 |
english_normalizationImproves number-reading performance in English text (dates, currencies, etc.). | Boolean | false | |
pronunciation_dictHow to say particular words, one "word/pronunciation" per entry (e.g. "Omg/Oh my god", or a phonetic spelling). | Text list | — | — |
Voices · 17
Use this model
await mindstudio.textToSpeech({ text: "...", speechModelOverride: { model: "minimax-speech-2.8-hd", config: { voice_id: "Wise_Woman", speed: 1, }, }, });