Introduction
MiniMax Speech 2.8 Turbo API is the speed-oriented speech option for previewing lines, revising scripts and producing many short recordings. Choose from 472 built-in voices for Turbo previews; keep the voice fixed while you compare revised lines, starting with Calm_Woman if unspecified. Turbo can detect the script language or use one of 40 language hints, useful when a brief line gives automatic detection little context. A Turbo preview arrives as a 32 kHz mono MP3 by default, with WAV, FLAC and raw PCM available for other playback pipelines. A Turbo request contains at most 5,000 characters, one system voice and one emotion; split separate speakers into separate tasks, with no cloning or word-level timing output. In Turbo scripts, punctuation sets the reading rhythm and supported parenthesized tags cue sounds; test the phrasing before duplicating a line across clips. Turbo batches are priced from submitted text at a rate per 1,000 characters, spaces and punctuation included; the quote is known before each run and failed tasks cost nothing.
| Field | Value |
|---|---|
| Input | Text |
| Output | Audio |
| Required field | text — 1 to 5,000 characters |
| Optional fields | Turbo — |
| Result | Turbo — 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 | The Turbo text field holds up to 5,000 characters per task; all submitted characters, including spaces, punctuation and performed tags such as (laughs), enter the quote. at least 1 characters · at most 5000 characters |
voice_id | No | Keep one of the 472 Turbo system voices fixed while revising text; |
emotion | No | Compare Turbo line deliveries with neutral, happy, sad, angry, fearful, disgusted or surprised; neutral is the default and each request keeps one emotion throughout. |
speed | No | Change Turbo pace from 0.5 to 2 to test whether a line fits a clip; the starting value is 1 and speed does not alter character billing. |
pitch | No | Try Turbo pitch shifts between -12 and 12 semitones while keeping the script fixed; 0 is unchanged and the experiment adds no surcharge. |
volume | No | Compare Turbo loudness using volume from 0.1 to 10, with 1 as the baseline; it changes the delivery level without changing the quoted price. |
language | No | Turbo uses auto by default; for a short line with ambiguous language, choose one of the 40 supported hints and listen again. |
output_format | No | Use mp3 for a normal Turbo preview, wav or flac for another file workflow, or pcm for headerless 16-bit little-endian samples; rate and channels apply to pcm, which needs wrappi… |
sample_rate | No | Turbo defaults to 32000 Hz; test 8000, 16000, 22050, 24000 or 44100 Hz when the destination requires it, without changing the character rate. |
bitrate | No | Turbo MP3 previews support 32000/64000/128000/256000 bits/s, default 128000; reducing bitrate reduces file size without reducing the synthesis price, and other formats ignore it. |
audio_channels | No | Turbo previews use mono unless you select stereo; keep this setting consistent across a comparison batch, at no extra cost. |
english_normalization | No | Turn on English normalization when a Turbo test misreads a number, date or abbreviation; initially off, with no effect on price. Default: |
pronunciation_dict | No | Keep Turbo names consistent with up to 100 {alias, pronunciation} substitutions, written as words or IPA in parentheses; every alias match changes, aliases cannot include a slas… at most 100 items |
Pricing
$0.072 / 1,000 characters
Turbo batches are priced from submitted text at a rate per 1,000 characters, spaces and punctuation included; the quote is known before each run and failed tasks cost nothing.
Use cases
Voice-overs for short videos
For short video drafts, generate Turbo narration early so you can check whether the rewritten line fits your edit.
Multilingual content
For a multilingual Turbo batch, write each script in its target language and test short lines with an explicit language hint.
Uncensored audio stories
For Turbo audio stories, review what was actually spoken and keep publication rights distinct from the ability to synthesize a line.
Emotional character reads
Compare the seven Turbo emotion settings on a short character line before committing to a larger batch.
Prompt guide & pro tips
Send the script in text
In Turbo scripts, punctuation sets the reading rhythm and supported parenthesized tags cue sounds; test the phrasing before duplicating a line across clips.
Pick a voice and emotion
Choose from 472 built-in voices for Turbo previews; keep the voice fixed while you compare revised lines, starting with Calm_Woman if unspecified.
Add a language hint if needed
Turbo can detect the script language or use one of 40 language hints, useful when a brief line gives automatic detection little context.
MiniMax Speech 2.8 Turbo API is the speed-oriented speech option for previewing lines, revising scripts and producing many short recordings. A Turbo preview arrives as a 32 kHz mono MP3 by default, with WAV, FLAC and raw PCM available for other playback pipelines. A Turbo request contains at most 5,000 characters, one system voice and one emotion; split separate speakers into separate tasks, with no cloning or word-level timing output.
Choose from 472 built-in voices for Turbo previews; keep the voice fixed while you compare revised lines, starting with Calm_Woman if unspecified. Turbo can detect the script language or use one of 40 language hints, useful when a brief line gives automatic detection little context. In Turbo scripts, punctuation sets the reading rhythm and supported parenthesized tags cue sounds; test the phrasing before duplicating a line across clips.
For short video drafts, generate Turbo narration early so you can check whether the rewritten line fits your edit. For a multilingual Turbo batch, write each script in its target language and test short lines with an explicit language hint. For Turbo audio stories, review what was actually spoken and keep publication rights distinct from the ability to synthesize a line. Compare the seven Turbo emotion settings on a short character line before committing to a larger batch.
Iterate a short Turbo line, keep the best script and settings, then submit the remaining batch as separate speech tasks. Turbo batches are priced from submitted text at a rate per 1,000 characters, spaces and punctuation included; the quote is known before each run and failed tasks cost nothing. Our SpicyAPI Turbo tests took about 8 seconds for roughly 300 characters; measure your own workload rather than treating that sample as an SLA. SpicyAPI does not add a filter to Turbo scripts; our tested passages were spoken without cuts, which is evidence about those runs rather than an unlimited promise. Save useful Turbo takes before the 14-day retention expires, and verify script permissions before distributing synthetic speech under an honest attribution.
SpicyAPI does not add a filter to Turbo scripts; our tested passages were spoken without cuts, which is evidence about those runs rather than an unlimited promise. Save useful Turbo takes before the 14-day retention expires, and verify script permissions before distributing synthetic speech under an honest attribution.
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: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "minimax/speech-2.8-turbo/text-to-speech",
"input": {
"text": "Welcome back. Your files finished rendering a minute ago and are ready to download.",
"emotion": "neutral",
"language": "auto",
"voice_id": "Calm_Woman",
"output_format": "mp3"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.




















