Introduction
MiniMax Speech 2.8 HD API turns a finished script into expressive narration, with fidelity as the priority rather than rapid iteration. For an HD narration, audition the 472 system voices on the same opening paragraph before recording the rest; Calm_Woman is the default. HD supports 40 language hints and auto detection; a language hint guides pronunciation rather than translating your narration. The HD deliverable defaults to a 32 kHz mono MP3; choose WAV or FLAC when your editing workflow needs another file format. One HD task accepts up to 5,000 characters and produces one voice with one overall emotion; it supplies neither word timings nor subtitles, and offers no voice cloning. For HD delivery, keep spoken words in text and use punctuation to shape pauses; sound tags such as (sighs) are performed as audio cues. HD narration is charged per 1,000 submitted characters, including spaces and punctuation, so the script determines the quote before synthesis; failed tasks are not charged.
| Field | Value |
|---|---|
| Input | Text |
| Output | Audio |
| Required field | text — 1 to 5,000 characters |
| Optional fields | HD — |
| Result | HD — One audio file: MP3 (32 kHz mono) by default, or WAV, FLAC, PCM |
| Billed quantity | Characters of text sent, per 1,000 |
| Typical time | About 8 seconds for 300 characters |
| Retention | Generated audio is kept for 14 days |
Request parameters
| Parameter | Required | Description |
|---|---|---|
text | Yes | Supply the final HD script in text, up to 5,000 characters; spaces, punctuation and sound tags count toward character billing, while cues such as (sighs) are performed. at least 1 characters · at most 5000 characters |
voice_id | No | For HD voice casting, select a |
emotion | No | Set the overall HD performance to neutral (default), happy, sad, angry, fearful, disgusted or surprised; this choice applies to the entire take. |
speed | No | HD speaking speed ranges from 0.5 to 2, with 1 preserving the base pace; audition sentence endings after adjusting it, without a price surcharge. |
pitch | No | Adjust the HD voice by -12 to 12 semitones; leave pitch at 0 when evaluating the original timbre, with no price change. |
volume | No | HD loudness uses a multiplier from 0.1 to 10, initially 1; adjust it for the take rather than rewriting the script, at the same price. |
language | No | Choose auto for HD language detection, or supply one of 40 language hints to guide pronunciation of the prepared script. |
output_format | No | HD exports mp3 by default, with wav, flac or pcm alternatives; pcm contains headerless 16-bit little-endian samples at your chosen rate and channel count, so use wav for browser… |
sample_rate | No | Match the HD edit to 8000, 16000, 22050, 24000, 32000 or 44100 Hz; 32000 is the default and selecting another rate does not change the charge. |
bitrate | No | For the HD MP3 export, choose 32000, 64000, 128000 or 256000 bits/s; the default is 128000 and this control has no effect on WAV, FLAC or PCM or on price. |
audio_channels | No | HD channel layout: mono is the default; select stereo to match a stereo editing project without changing the charge. |
english_normalization | No | Enable this for HD English narration that needs numbers, dates and abbreviations expanded for speaking; it is off initially and adds no charge. Default: |
pronunciation_dict | No | For HD names or specialist terms, add up to 100 {alias, pronunciation} entries using words or parenthesized IPA; aliases cannot contain slashes, every match is replaced, and onl… at most 100 items |
Pricing
$0.12 / 1,000 characters
HD narration is charged per 1,000 submitted characters, including spaces and punctuation, so the script determines the quote before synthesis; failed tasks are not charged.
Use cases
Voice-overs for short videos
Listen for names, phrasing and sentence endings in your HD voice-over before placing it under the final video.
Multilingual content
For localized HD narration, prepare the script in the target language and check difficult names with pronunciation_dict.
Uncensored audio stories
For an audio story made with HD, review the entire performance and obtain the necessary script rights before publication.
Emotional character reads
Build an HD character performance around one of seven emotions, then listen for consistency across the whole passage.
Prompt guide & pro tips
Send the script in text
For HD delivery, keep spoken words in text and use punctuation to shape pauses; sound tags such as (sighs) are performed as audio cues.
Pick a voice and emotion
For an HD narration, audition the 472 system voices on the same opening paragraph before recording the rest; Calm_Woman is the default.
Add a language hint if needed
HD supports 40 language hints and auto detection; a language hint guides pronunciation rather than translating your narration.
MiniMax Speech 2.8 HD API turns a finished script into expressive narration, with fidelity as the priority rather than rapid iteration. The HD deliverable defaults to a 32 kHz mono MP3; choose WAV or FLAC when your editing workflow needs another file format. One HD task accepts up to 5,000 characters and produces one voice with one overall emotion; it supplies neither word timings nor subtitles, and offers no voice cloning.
For an HD narration, audition the 472 system voices on the same opening paragraph before recording the rest; Calm_Woman is the default. HD supports 40 language hints and auto detection; a language hint guides pronunciation rather than translating your narration. For HD delivery, keep spoken words in text and use punctuation to shape pauses; sound tags such as (sighs) are performed as audio cues.
Listen for names, phrasing and sentence endings in your HD voice-over before placing it under the final video. For localized HD narration, prepare the script in the target language and check difficult names with pronunciation_dict. For an audio story made with HD, review the entire performance and obtain the necessary script rights before publication. Build an HD character performance around one of seven emotions, then listen for consistency across the whole passage.
Generate a short HD sample first, review the chosen voice and pronunciation, then synthesize the remaining script with those settings. HD narration is charged per 1,000 submitted characters, including spaces and punctuation, so the script determines the quote before synthesis; failed tasks are not charged. SpicyAPI measured about 8 seconds for an HD passage of roughly 300 characters; this observation is not a delivery deadline. SpicyAPI adds no script filter to HD; our existing tests read the submitted lines in full, but those observations do not guarantee every future performance. Download an HD recording within 14 days, check your rights to the script, and identify synthetic narration honestly rather than attributing it to a real speaker.
SpicyAPI adds no script filter to HD; our existing tests read the submitted lines in full, but those observations do not guarantee every future performance. Download an HD recording within 14 days, check your rights to the script, and identify synthetic narration honestly rather than attributing it to a real speaker.
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "minimax/speech-2.8-hd/text-to-speech",
"input": {
"text": "Welcome back. Your files finished rendering a minute ago and are ready to download.",
"emotion": "neutral",
"language": "auto",
"voice_id": "Calm_Woman",
"output_format": "mp3"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.




















