MiniMax Speech 01 HD converts supplied text into spoken audio. Its exact Ailoot model ID is minimax/speech-01-hd, and its documented synthesis selector is speech-01-hd. Use the audio_speech method for a voiceover, spoken lesson, product explanation or narration draft whose wording you already have. The input is the passage to read, and the result is synthesized speech in the selected audio format.
Prepare the text as words intended to be spoken, with punctuation that reflects the phrasing. For a narration draft, supply the final sentence order and spellings of names before choosing the voice settings. Keep stage directions separate from the spoken passage unless those words belong in the recording. Listen to important names, numbers and transitions in the result and compare them with the source text.
The text must be shorter than 10,000 characters. The published synthesis contract requires model and text. For longer material, prepare passages within that limit and retain their order for editing. The synchronous profile uses a complete response, with stream=false as the documented default. The general API also describes streaming options, but this card's selected method is the synchronous speech workflow.
voice_setting selects an available voice_id and controls speed, vol and pitch. The documented speed range is 0.5–2, with 1 as the default. vol must be greater than 0 and at most 10, with 1 as the default; pitch is an integer from -12 to 12, with 0 as the default. Make each setting serve the passage: compare the reading pace and audibility of the actual recording rather than assuming a numeric change creates a particular performance.
The documented emotion values are happy, sad, angry, fearful, disgusted, surprised, calm and fluent. The text can also guide the model's emotion. text_normalization is a separate boolean control for Chinese and English number-reading preparation, and latex_read controls reading LaTeX formulas; both default to false. pronunciation_dict.tone can carry documented pronunciation replacement rules. Check a name or specialized expression in the resulting audio before using it in a longer narration.
audio_setting.format accepts mp3, pcm, flac or wav, with mp3 as the default. sample_rate accepts 8000, 16000, 22050, 24000, 32000 or 44100 and defaults to 32000. bitrate accepts 32000, 64000, 128000 or 256000, defaults to 128000 and applies to MP3. channel selects 1 for mono or 2 for stereo, with 1 as the default. Keep these choices with the source passage so the returned file can be checked for the intended editing or playback destination.
Additional documented controls include language_boost, whose exact named choices include English, Russian, Chinese and Chinese,Yue; choose an actual listed value for the passage. timber_weights can combine existing voice_id entries with integer weight values from 1 to 100. The published sound_effects values are spacious_echo, auditorium_echo, lofi_telephone and robotic. These are optional controls for shaping a selected voice; record their actual settings when comparing readings.
The synchronous result contains generated speech. output_format selects url or hex and defaults to hex in the synthesis schema. Download the file referenced by a URL result, or decode a hex result into audio bytes before playback. Keep the returned audio metadata, including its format, sample rate, channel count and size, with the recording. subtitle_enable is false by default and applies only to non-streaming output when enabled; the source does not establish a subtitle file format for this profile.
Save the source passage, exact model ID, voice selection and completed recording together. Review the whole reading for intended wording, pauses and the chosen delivery settings before placing it in a video, lesson or spoken guide. If the response is incomplete, keep its original request identity and returned information for reconciliation before requesting another recording. A later revision can change a specific phrase or voice setting while preserving the rest of the script for a clear comparison.
", "aboutText": "MiniMax Speech 01 HD converts supplied text into spoken audio. Its exact Ailoot model ID is minimax/speech-01-hd, and its documented synthesis selector is speech-01-hd. Use the audio_speech method for a voiceover, spoken lesson, product explanation or narration draft whose wording you already have. The input is the passage to read, and the result is synthesized speech in the selected audio format.\n\nPrepare the text as words intended to be spoken, with punctuation that reflects the phrasing. For a narration draft, supply the final sentence order and spellings of names before choosing the voice settings. Keep stage directions separate from the spoken passage unless those words belong in the recording. Listen to important names, numbers and transitions in the result and compare them with the source text.\n\nThe text must be shorter than 10,000 characters. The published synthesis contract requires model and text. For longer material, prepare passages within that limit and retain their order for editing. The synchronous profile uses a complete response, with stream=false as the documented default. The general API also describes streaming options, but this card's selected method is the synchronous speech workflow.\n\nvoice_setting selects an available voice_id and controls speed, vol and pitch. The documented speed range is 0.5–2, with 1 as the default. vol must be greater than 0 and at most 10, with 1 as the default; pitch is an integer from -12 to 12, with 0 as the default. Make each setting serve the passage: compare the reading pace and audibility of the actual recording rather than assuming a numeric change creates a particular performance.\n\nThe documented emotion values are happy, sad, angry, fearful, disgusted, surprised, calm and fluent. The text can also guide the model's emotion. text_normalization is a separate boolean control for Chinese and English number-reading preparation, and latex_read controls reading LaTeX formulas; both default to false. pronunciation_dict.tone can carry documented pronunciation replacement rules. Check a name or specialized expression in the resulting audio before using it in a longer narration.\n\naudio_setting.format accepts mp3, pcm, flac or wav, with mp3 as the default. sample_rate accepts 8000, 16000, 22050, 24000, 32000 or 44100 and defaults to 32000. bitrate accepts 32000, 64000, 128000 or 256000, defaults to 128000 and applies to MP3. channel selects 1 for mono or 2 for stereo, with 1 as the default. Keep these choices with the source passage so the returned file can be checked for the intended editing or playback destination.\n\nAdditional documented controls include language_boost, whose exact named choices include English, Russian, Chinese and Chinese,Yue; choose an actual listed value for the passage. timber_weights can combine existing voice_id entries with integer weight values from 1 to 100. The published sound_effects values are spacious_echo, auditorium_echo, lofi_telephone and robotic. These are optional controls for shaping a selected voice; record their actual settings when comparing readings.\n\nThe synchronous result contains generated speech. output_format selects url or hex and defaults to hex in the synthesis schema. Download the file referenced by a URL result, or decode a hex result into audio bytes before playback. Keep the returned audio metadata, including its format, sample rate, channel count and size, with the recording. subtitle_enable is false by default and applies only to non-streaming output when enabled; the source does not establish a subtitle file format for this profile.\n\nSave the source passage, exact model ID, voice selection and completed recording together. Review the whole reading for intended wording, pauses and the chosen delivery settings before placing it in a video, lesson or spoken guide. If the response is incomplete, keep its original request identity and returned information for reconciliation before requesting another recording. A later revision can change a specific phrase or voice setting while preserving the rest of the script for a clear comparison.", "editorialOverlay": { "copy_fields": null, "copy_sha256": null, "copy_source": null, "html_sha256": "e57641d44322594c49e82525bbf2f6d3349762751586627dd0f4db8ddc2cee9a", "module_sha256": "ee9dc1c50b1e6a8fb7fa062030fcb4646b9aaaeb08be513b723a5a92a7290f06", "raw_model_unchanged": true, "schema": "ailoot-native-model-editorial-v1" }, "locale": "en", "methods": [ { "html": "minimax/speech-01-hdMiniMax Speech 01 HD reads supplied text as generated speech through a synchronous audio_speech request. It accepts text shorter than 10,000 characters and offers voice, pace and pitch settings with MP3, PCM, FLAC or WAV output.
This description covers the model’s capabilities. Available in Ailoot inputs, outputs and options are listed for the selected method below.
MiniMax Speech 01 HD converts supplied text into spoken audio. Its exact Ailoot model ID is minimax/speech-01-hd, and its documented synthesis selector is speech-01-hd. Use the audio_speech method for a voiceover, spoken lesson, product explanation or narration draft whose wording you already have. The input is the passage to read, and the result is synthesized speech in the selected audio format.
Prepare the text as words intended to be spoken, with punctuation that reflects the phrasing. For a narration draft, supply the final sentence order and spellings of names before choosing the voice settings. Keep stage directions separate from the spoken passage unless those words belong in the recording. Listen to important names, numbers and transitions in the result and compare them with the source text.
The text must be shorter than 10,000 characters. The published synthesis contract requires model and text. For longer material, prepare passages within that limit and retain their order for editing. The synchronous profile uses a complete response, with stream=false as the documented default. The general API also describes streaming options, but this card's selected method is the synchronous speech workflow.
voice_setting selects an available voice_id and controls speed, vol and pitch. The documented speed range is 0.5–2, with 1 as the default. vol must be greater than 0 and at most 10, with 1 as the default; pitch is an integer from -12 to 12, with 0 as the default. Make each setting serve the passage: compare the reading pace and audibility of the actual recording rather than assuming a numeric change creates a particular performance.
The documented emotion values are happy, sad, angry, fearful, disgusted, surprised, calm and fluent. The text can also guide the model's emotion. text_normalization is a separate boolean control for Chinese and English number-reading preparation, and latex_read controls reading LaTeX formulas; both default to false. pronunciation_dict.tone can carry documented pronunciation replacement rules. Check a name or specialized expression in the resulting audio before using it in a longer narration.
audio_setting.format accepts mp3, pcm, flac or wav, with mp3 as the default. sample_rate accepts 8000, 16000, 22050, 24000, 32000 or 44100 and defaults to 32000. bitrate accepts 32000, 64000, 128000 or 256000, defaults to 128000 and applies to MP3. channel selects 1 for mono or 2 for stereo, with 1 as the default. Keep these choices with the source passage so the returned file can be checked for the intended editing or playback destination.
Additional documented controls include language_boost, whose exact named choices include English, Russian, Chinese and Chinese,Yue; choose an actual listed value for the passage. timber_weights can combine existing voice_id entries with integer weight values from 1 to 100. The published sound_effects values are spacious_echo, auditorium_echo, lofi_telephone and robotic. These are optional controls for shaping a selected voice; record their actual settings when comparing readings.
The synchronous result contains generated speech. output_format selects url or hex and defaults to hex in the synthesis schema. Download the file referenced by a URL result, or decode a hex result into audio bytes before playback. Keep the returned audio metadata, including its format, sample rate, channel count and size, with the recording. subtitle_enable is false by default and applies only to non-streaming output when enabled; the source does not establish a subtitle file format for this profile.
Save the source passage, exact model ID, voice selection and completed recording together. Review the whole reading for intended wording, pauses and the chosen delivery settings before placing it in a video, lesson or spoken guide. If the response is incomplete, keep its original request identity and returned information for reconciliation before requesting another recording. A later revision can change a specific phrase or voice setting while preserving the rest of the script for a clear comparison.
POST /v1/audio/speechtextaudio| Field | Type | Requirement | Values & limits | Default |
|---|---|---|---|---|
Fieldmodel | Typestring | RequirementRequired | Values & limits minimax/speech-01-hd | Default— |
Fieldtext | Typestring | RequirementRequired | Values & limits | Default— |
Fieldstream | Typeboolean | RequirementOptional | Values & limits false | Defaultfalse |
Fieldvoice_setting | Typeobject | RequirementOptional | Values & limits | Default— |
Fieldvoice_setting.voice_id | Typestring | RequirementOptional | Values & limits | Default— |
Fieldvoice_setting.speed | Typenumber | RequirementOptional | Values & limits Minimum: 0.5 Maximum: 2 | Default1 |
Fieldvoice_setting.vol | Typenumber | RequirementOptional | Values & limits Minimum: 0 Maximum: 10 | Default1 |
Fieldvoice_setting.pitch | Typeinteger | RequirementOptional | Values & limits Minimum: -12 Maximum: 12 | Default0 |
Fieldvoice_setting.emotion | Typestring | RequirementOptional | Values & limits happy, sad, angry, fearful, disgusted, surprised, calm, fluent | Default— |
Fieldvoice_setting.text_normalization | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldvoice_setting.latex_read | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldaudio_setting | Typeobject | RequirementOptional | Values & limits | Default— |
Fieldaudio_setting.sample_rate | Typeinteger | RequirementOptional | Values & limits 8,000, 16,000, 22,050, 24,000, 32,000, 44,100 | Default32,000 |
Fieldaudio_setting.bitrate | Typeinteger | RequirementOptional | Values & limits 32,000, 64,000, 128,000, 256,000 | Default128,000 |
Fieldaudio_setting.format | Typestring | RequirementOptional | Values & limits mp3, pcm, flac, wav | Defaultmp3 |
Fieldaudio_setting.channel | Typeinteger | RequirementOptional | Values & limits 1, 2 | Default1 |
Fieldaudio_setting.force_cbr | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldpronunciation_dict | Typeobject | RequirementOptional | Values & limits | Default— |
Fieldpronunciation_dict.tone | Typearray | RequirementOptional | Values & limits | Default— |
Fieldtimber_weights | Typearray | RequirementOptional | Values & limits | Default— |
Fieldlanguage_boost | Typestring | RequirementOptional | Values & limits 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— |
Fieldvoice_modify | Typeobject | RequirementOptional | Values & limits | Default— |
Fieldvoice_modify.pitch | Typeinteger | RequirementOptional | Values & limits -100, 100 | Default— |
Fieldvoice_modify.intensity | Typeinteger | RequirementOptional | Values & limits -100, 100 | Default— |
Fieldvoice_modify.timbre | Typeinteger | RequirementOptional | Values & limits -100, 100 | Default— |
Fieldvoice_modify.sound_effects | Typestring | RequirementOptional | Values & limits spacious_echo, auditorium_echo, lofi_telephone, robotic | Default— |
Fieldsubtitle_enable | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
Fieldoutput_format | Typestring | RequirementOptional | Values & limits url, hex | Defaulthex |
Fieldaigc_watermark | Typeboolean | RequirementOptional | Values & limits | Defaultfalse |
| Parameter | Contract |
|---|---|
Parametercontent_type | Contractapplication/json |
Parameterrequired | Contractmodel, text |
Parameteroptional | Contractstream, voice_setting, audio_setting, pronunciation_dict, timber_weights, language_boost, voice_modify, subtitle_enable, output_format, aigc_watermark |
Parametermaterial_types | Contracttext |
Parameterconditions | ContractUse text, not input or contents. Fewer than 10000 Unicode characters; stream is false or omitted., PCM is raw audio with its returned sample geometry; do not invent a WAV header or bit depth. |
Lifecycle: synchronous_result_then_owned_library
GET /v1/media/assets
GET /v1/media/assets/{asset_id}
Tariff references do not add request options. Supported parameters are listed for the selected method.
Match each charge to the conditions shown. Use one rate for each set of alternative configurations.
| Configuration | Rate schedule | Billing unit | Charge |
|---|---|---|---|
| ConfigurationBase token rates | Rate scheduleBase rate | Billing unit— | Charge Input · Input tokens — / 1K tokens Output · Output tokens — / 1K tokens Input · Cache read — / 1K tokens Input · Cache write — / 1K tokens Input · Cache write, 1 hour — / 1K tokens |
| Configurationinput character | Rate schedule— | Billing unitper input character | Charge Input · input character $0.000105 per input character |
Ailoot reserves an estimated amount before a request. After the result, it settles the final charge from measured usage. Streaming settlement uses the final usage record. How reservation and settlement work →
set -eu\n: "${AILOOT_API_KEY:?Set your Ailoot API key}"\nexport AILOOT_API_KEY\npython3 - <<'PY_REQUEST'\nimport json,os\nfrom pathlib import Path\np=json.loads('{"model": "minimax/speech-01-hd", "text": "Hello, this is a short speech example.", "stream": false, "output_format": "hex", "audio_setting": {"format": "mp3", "sample_rate": 32000, "channel": 1}}')\n\nPath("request.json").write_text(json.dumps(p,ensure_ascii=False))\nPY_REQUEST\ncurl --fail-with-body --silent --show-error https://api.ailoot.ai/v1/audio/speech \\\n -H "Authorization: Bearer $AILOOT_API_KEY" -H 'Content-Type: application/json' \\\n --data-binary @request.json -D response-headers.txt -o result.json\n# Submit once. Retain the complete result, original identifiers and every response header.\nmp3,pcm,flac,wav
GET /v1/media/assetsGET /v1/media/assets/{asset_id}Keep every complete returned response, file and content part.
Retain the full JSON. Decode every hex audio byte or use the returned owned URL. Keep the actual format, sample rate, channel count and usage_characters.
Preserve the original response, headers and X-Client-Request-ID. An uncertain outcome does not authorize another creation.
| Field | Type |
|---|---|
Fielddata.audio | Typestring |
Fieldextra_info | Typeobject |
Fieldtrace_id | Typestring |
Ailoot Ultra minimax/speech-01-hd
Use the method paths shown above.