Minimax Speech 2.8 HD

minimax_speech_28_hd

Minimax Speech 2.8 HD converts text into speech audio and supports voice, audio, subtitle, pronunciation, and voice-effect settings.

Эндпоинты

POST /v1/queue/minimax_speech_28_hd # Создать задачу

GET /v1/queue/minimax_speech_28_hd/requests/{request_id}/status # Проверить статус

GET /v1/queue/minimax_speech_28_hd/requests/{request_id}/response # Получить результат

PUT /v1/queue/minimax_speech_28_hd/requests/{request_id}/cancel # Отменить задачу

Параметры запроса

curl --location "https://odirouter.ai/model/v1/queue/minimax_speech_28_hd" \
  --header "Authorization: Bearer YOUR_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "text": "Hello, welcome to OdiRouter Model API.",
    "output_format": "url",
    "voice_setting": {
      "voice_id": "female-shaonv",
      "speed": 1.08,
      "vol": 1,
      "pitch": 0
    },
    "audio_setting": {
      "sample_rate": 32000,
      "bitrate": 128000,
      "channel": 1,
      "format": "mp3"
    },
    "language_boost": "English"
  }'

text string обязательно

Text to synthesize into speech. The length must be less than 10000 characters. Streaming output is recommended when the text is longer than 3000 characters.

Use newline characters to separate paragraphs.

Pause control: add <#x#> in the text, where x is the pause duration in seconds. Range: 0.01 - 99.99. Keep at most two decimal places. Pause markers must be placed between pronounceable text segments and cannot be used continuously.

Interjection tags are supported, including (laughs), (chuckle), (coughs), (clear-throat), (groans), (breath), (pant), (inhale), (exhale), (gasps), (sniffs), (sighs), (snorts), (burps), (lip-smacking), (humming), (hissing), (emm), (sneezes).

output_format string

Audio output format.

Options: url, hex

Default: url

voice_setting object

Voice settings, including voice ID, speed, volume, and pitch.

voice_id string

Voice ID used for speech synthesis. To use mixed voices, set timbre_weights and leave this field empty.

System voices, cloned voices, and text-generated voices are supported. See the system voice list for system voice IDs.

speed number

Speech speed. Larger values produce faster speech.

Range: 0.5 - 2

Default: 1.0

vol number

Speech volume. Larger values produce louder audio.

Range: 0 - 10

Default: 1.0

pitch integer

Speech pitch.

Range: -12 - 12

Default: 0

emotion string

Controls the emotion of synthesized speech. The model usually selects a suitable emotion based on the input text, so manual configuration is generally unnecessary.

Options: happy, sad, angry, fearful, disgusted, surprised, calm, fluent

text_normalization boolean

Whether to enable Chinese and English text normalization. This can improve number-reading performance but may slightly increase latency.

Default: false

latex_read boolean

Whether to read LaTeX formulas aloud. This only supports Chinese and sets language_boost to Chinese when enabled.

Wrap formulas with $$. If the formula contains \, escape it as \\, for example $$x = \\frac{-b \\pm \\sqrt{b^2 - 4ac}}{2a}$$.

Default: false

timbre_weights array

Mixed voice configuration. Up to 4 voices can be mixed. Each item contains the following fields.

voice_id string

Voice ID used for speech synthesis. This must be filled together with weight. System voices, cloned voices, and text-generated voices are supported. See the system voice list for system voice IDs.

weight integer

Weight of each voice in the synthesized audio. This must be filled together with voice_id. The higher a single voice's weight, the more similar the synthesized voice is to that voice.

Range: 1 - 100

audio_setting object

Audio settings.

sample_rate integer

Sample rate of the generated audio.

Options: 8000, 16000, 22050, 24000, 32000, 44100

Default: 32000

bitrate integer

Bitrate of the generated audio. This parameter only takes effect for mp3 audio.

Options: 32000, 64000, 128000, 256000

Pass this as a JSON number, for example 128000, not a quoted string.

Default: 128000

format string

Generated audio format. wav is only supported for non-streaming output.

Options: mp3, pcm, flac, wav

channel integer

Number of generated audio channels.

Options: 1, 2

Default: 1

force_cbr boolean

Constant bitrate control. When set to true, audio is encoded with constant bitrate. This only takes effect for streaming output with mp3 audio.

Default: false

language_boost string

Whether to enhance recognition for a specified language or dialect. Use auto to let the model choose automatically.

Options: Chinese, Chinese,Yue, English, Arabic, Russian, Spanish, French, Portuguese, German, Turkish, Dutch, Ukrainian, Vietnamese, Indonesian, Japanese, Italian, Korean, Thai, Polish, Romanian, Greek, Czech, Finnish, Hindi, Bulgarian, Danish, Hebrew, Malay, Persian, Slovak, Swedish, Croatian, Filipino, Hungarian, Norwegian, Slovenian, Catalan, Nynorsk, Tamil, Afrikaans, auto

Default: null

pronunciation_dict object

Pronunciation dictionary used to customize pronunciations for specific words.

tone array

Defines pronunciation replacement rules for marked text or symbols. In Chinese text, tones are represented by numbers: first tone is 1, second tone is 2, third tone is 3, fourth tone is 4, and neutral tone is 5.

Example: ["燕少飞/(yan4)(shao3)(fei1)", "omg/oh my god"]

voice_modify object

Voice effect settings. Supported audio formats: non-streaming mp3, wav, flac; streaming mp3.

pitch integer

Pitch adjustment. Values near -100 produce a lower voice; values near 100 produce a brighter voice.

Range: -100 - 100

intensity integer

Intensity adjustment. Values near -100 produce a stronger voice; values near 100 produce a softer voice.

Range: -100 - 100

timbre integer

Timbre adjustment. Values near -100 produce a richer voice; values near 100 produce a clearer voice.

Range: -100 - 100

sound_effects string

Sound effect setting. Only one effect can be selected per request.

Options: spacious_echo (spacious echo), auditorium_echo (auditorium broadcast), lofi_telephone (telephone distortion), robotic (electronic voice)

subtitle_enable bool

Whether to enable subtitle generation. This parameter only takes effect for non-streaming output.

Default: false

subtitle_type string

Subtitle granularity.

Options: sentence (sentence-level timestamps), word (word-level timestamps), word_streaming (streaming-optimized word-level timestamps; only applies to upstream streaming scenarios; OdiRouter queue does not expose a streaming control field)

Default: sentence

aigc_watermark bool

Whether to append an audio rhythm marker to the end of synthesized audio. This only takes effect for non-streaming synthesis.

Default: false

Параметры ответа

{
  "request_id": "mmsu_01hxyz...",
  "response_url": "https://odirouter.ai/model/v1/queue/minimax_speech_28_hd/requests/mmsu_01hxyz.../response",
  "status_url": "https://odirouter.ai/model/v1/queue/minimax_speech_28_hd/requests/mmsu_01hxyz.../status",
  "cancel_url": "https://odirouter.ai/model/v1/queue/minimax_speech_28_hd/requests/mmsu_01hxyz.../cancel",
  "queue_position": 0
}

Поле	Тип	Описание
`request_id`	string	Идентификатор задачи, используемый во всех последующих операциях.
`response_url`	string	URL для получения итогового результата.
`status_url`	string	URL для проверки статуса задачи.
`cancel_url`	string	URL для отмены задачи.
`queue_position`	int	Текущая позиция в очереди.

Следующие операции