# Novita AI MiniMax Speech 2.5 HD Preview API | High-Definition Text-to-Speech > For the complete documentation index, see [llms.txt](/llms.txt). Markdown is available with `Accept: text/markdown` and `.md` URL variants. Source: /docs/api-reference/model-apis-minimax-speech-2.5-hd-preview # MiniMax Speech-2.5-hd-preview Text to Speech POST / v3 / minimax-speech-2.5-hd-preview Try it MiniMax Speech-2.5-hd-preview Text to Speech cURL ``` curl --request POST \ --url https://api.novita.ai/v3/minimax-speech-2.5-hd-preview \ --header 'Authorization: <authorization>' \ --header 'Content-Type: <content-type>' \ --data ' { "text": "<string>", "voice_setting": { "speed": 123, "vol": 123, "pitch": 123, "voice_id": "<string>", "emotion": "<string>", "text_normalization": true }, "audio_setting": { "sample_rate": 123, "bitrate": 123, "format": "<string>", "channel": 123 }, "pronunciation_dict": { "tone": [ {} ] }, "timbre_weights": [ { "voice_id": "<string>", "weight": 123 } ], "stream": true, "language_boost": "<string>", "output_format": "<string>", "voice_modify": { "pitch": 123, "intensity": 123, "timbre": 123, "sound_effects": "<string>" } } ' ``` 200 ``` { "audio": "<string>", "status": 123 } ``` MiniMax’s high-definition text-to-speech model, Compared to Speech 02 released in May, Speech 2.5 has three major breakthroughs: stronger multilingual expressiveness, more accurate voice replication, and broader coverage with 40 languages. Best suited for short sentence generation, voice chat, and online social scenarios. Processing is fast, but the text length is limited to less than 10,000 characters. For long-form text, we recommend using [asynchronous TTS synthesis](/docs/api-reference/model-apis-minimax-speech-2.5-hd-preview-async). ## [​](#request-headers) Request Headers [​](#param-content-type) Content-Type string required Supports: `application/json` [​](#param-authorization) Authorization string required Bearer authentication format, for example: Bearer {{API Key}}. ## [​](#request-body) Request Body [​](#param-text) text string required The text to be synthesized. The length must be less than 10,000 characters. Use line breaks to separate paragraphs. To control the pause duration between speech segments, insert `<#x#>` between words or sentences, where `x` is the pause duration in seconds (supports 0.01–99.99, up to two decimal places). Custom speech pauses between text segments are supported, allowing you to control the timing of pauses in the generated audio. Note: Pause markers must be placed between two segments that can be pronounced, and multiple consecutive pause markers are not allowed. [​](#param-voice-setting) voice_setting object required Show properties [​](#param-speed) speed float default:"1.0" Range: [0.5, 2], default is 1.0Controls the speech rate of the generated audio. Optional. Higher values result in faster speech. [​](#param-vol) vol float default:"1.0" Range: (0, 10], default is 1.0Controls the volume of the generated audio. Optional. Higher values result in louder audio. [​](#param-pitch) pitch int default:"0" Range: [-12, 12], default is 0Controls the pitch of the generated audio. Optional. 0 means original timbre. The value must be an integer. [​](#param-voice-id) voice_id string The requested timbre ID. Required (choose either this or timbre_weights).Supports both system voices (ID) and cloned voices (ID). The available system voice IDs are as follows: - Wise_Woman - Friendly_Person - Inspirational_girl - Deep_Voice_Man - Calm_Woman - Casual_Guy - Lively_Girl - Patient_Man - Young_Knight - Determined_Man - Lovely_Girl - Decent_Boy - Imposing_Manner - Elegant_Man - Abbess - Sweet_Girl_2 - Exuberant_Girl [​](#param-emotion) emotion string Controls the emotion of the synthesized speech.Currently supports 7 emotions: happy, sad, angry, fearful, disgusted, surprised, neutral.Allowed values: `["happy", "sad", "angry", "fearful", "disgusted", "surprised", "neutral"]` [​](#param-text-normalization) text_normalization bool default:"false" This parameter supports English text normalization, which improves performance in number-reading scenarios, but this comes at the cost of a slight increase in latency. If not provided, the default is false. [​](#param-audio-setting) audio_setting object Show properties [​](#param-sample-rate) sample_rate number default:32000 Range: [8000, 16000, 22050, 24000, 32000, 44100]The sample rate of the generated audio. Optional, default is 32000. [​](#param-bitrate) bitrate number default:128000 Range: [32000, 64000, 128000, 256000]The bitrate of the generated audio. Optional, default is 128000. This parameter only applies to mp3 audio format. [​](#param-format) format string default:"mp3" The audio format of the output. Default is mp3. Options: `mp3`, `pcm`, `flac`, `wav`. wav is only supported for non-streaming output. [​](#param-channel) channel number default:1 Number of audio channels. Default is 1 (mono). Options:1: mono2: stereo [​](#param-pronunciation-dict) pronunciation_dict object Show properties [​](#param-tone) tone list Replacement of text, symbols and corresponding pronunciations that require manual handling. Replace the pronunciation (adjust the tone/replace the pronunciation of other characters) using the following format:[“omg/oh my god”]For Chinese texts, tones are replaced by numbers, with 1 for the first tone (high), 2 for the second tone (rising), 3 for the third tone (low/dipping), 4 for the fourth tone (falling), and 5 for the fifth tone (neutral). [​](#param-timbre-weights) timbre_weights object[] Required if voice_id is not provided (choose one of the two). Show properties [​](#param-voice-id-1) voice_id string The requested timbre ID. Must be provided together with the weight parameter. [​](#param-weight) weight number Range: [1, 100]The weight, must be provided together with voice_id. Up to 4 timbres can be mixed. The value must be an integer. The higher the proportion for a single timbre, the more the synthesized voice will resemble it. [​](#param-stream) stream boolean default:"false" Whether to enable streaming. Default is false, i.e., streaming is disabled. [​](#param-language-boost) language_boost string default:"null" Enhances recognition of specified minor languages and dialects. Setting this parameter can improve speech performance in the specified language/dialect scenarios. If the minor language type is not clear, you can set it to “auto” and the model will automatically determine the language type. Supported values:`'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'` [​](#param-output-format) output_format string default:"hex" Controls the output format of the result. Optional values are `url` and `hex`. The default is `hex`. This parameter only takes effect in non-streaming scenarios; in streaming mode, only hex format is supported. The returned URL is valid for 24 hours. [​](#param-voice-modify) voice_modify object Voice FX settings. Supported audio formats for this parameter: - Non-streaming: mp3, wav, flac - Streaming: mp3 Show properties [​](#param-pitch-1) pitch integer Pitch adjustment (darker/brighter). Range: [-100, 100]. Values closer to -100 generate a deeper (darker) voice; values closer to 100 produce a brighter voice. [​](#param-intensity) intensity integer Intensity adjustment (powerful/soft). Range: [-100, 100]. Values closer to -100 generate a more powerful sound; values closer to 100 result in a softer sound. [​](#param-timbre) timbre integer Timbre adjustment (magnetic/crisp). Range: [-100, 100]. Values closer to -100 make the voice more full/magnetic; values closer to 100 make it crisper. [​](#param-sound-effects) sound_effects string Sound effect setting (only one may be selected per request). Valid values: - `spacious_echo` (large space echo) - `auditorium_echo` (auditorium broadcast) - `lofi_telephone` (telephone distortion) - `robotic` (robotic effect) ## [​](#response) Response [​](#param-audio) audio string The synthesized audio segment, encoded in hex and generated in the format specified by `audio_setting.format` (mp3/pcm/flac). The return format is determined by the `output_format` parameter. When `stream` is true, only hex format is supported. [​](#param-status) status number The current status of the audio stream. Returned only when `stream` is true. 1 indicates synthesis in progress, 2 indicates synthesis completed. Last modified on February 5, 2026