Minimax Speech 2.6 Hd API Documentation

Minimax Speech 2.6 Hd API Documentation

Playground

Minimax Speech 2.6 HD: Ultra-human, low-latency (< 250ms) TTS with voice cloning, text normalization and support for 40+ languages. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

High-definition Text-to-Speech (TTS) with natural pronunciation and crisp articulation. Supports custom cloned voices and built-in voices, adjustable speed, volume, and pitch, and coverage of 40+ languages for professional audio creation.

Features

• Multilingual upgrade: Stronger English and overall multilingual similarity, accuracy, and rhythm vs. Speech 02; seamless switching across 40+ languages for meetings, podcasts, and daily dialog.
• Lifelike tone replication: Control across language, accent, style, and emotion—preserves cross-language and regional accents and “age” timbre with high fidelity.
• Global language set (40+): Expanded catalog including Bulgarian, Danish, Hebrew, Malay, Persian, Slovak, Swedish, Croatian, Filipino, Hungarian, Norwegian, Slovenian, Catalan, Nynorsk, Tamil, Afrikaans, and more—ideal for cross-border commerce, customer support, and localized marketing.
• Real-time streaming capabilities: generate and play audio as it’s being synthesized, enabling low-latency experiences for live or interactive applications.

How to Use

1) Choose a Voice (voice_id)

Use either a custom voice you trained (voice cloning) or a built-in system voice (case-sensitive):

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

• See the full, list and samples: Voice_ID list

2) Set Audio Parameters (mapped to the UI dropdowns)

• english_normalization (boolean) Improves English text normalization, especially numbers/units (e.g., “$1,299” → “one thousand two hundred ninety-nine dollars”).
• sample_rate (Hz) 22050 / 24000 / 44100 / 48000. Tip: 44.1 kHz for music/podcasts; 48 kHz for video.
• bitrate (bps for MP3/OGG) 64k / 96k / 128k / 192k / 256k / 320k. Tip: ≥192k for distribution; 96–128k for previews.
• channel: mono or stereo Mono is smaller/clearer for speech; stereo for spatial mixes.
• format: mp3, wav, ogg, flac, wav is lossless (larger); mp3 is compact and web-friendly. Also supports streaming PCM output for real-time playback.
• language_boost (IETF code: en, zh, ja, …) Prioritizes a primary language in mixed-language inputs.

Prosody controls

• speed: speaking rate (e.g., 0.8–1.2)
• volume: gain (linear or dB, depending on API)
• pitch: pitch shift (semitones/cents or normalized)

Pricing

• Price: $0.10 / 1,000 characters

Quick examples

• 1,000 chars → $0.10
• 2,500 chars → $0.10 × 2.5 = $0.25
• 10,000 chars → $1.00

Typical Use Cases

Short-video and ad voiceovers, e-learning/courseware, AI assistants and IVR, podcasts/audiobooks, cross-border e-commerce localization.

Best-Practice Presets (optional)

• Video voiceover: format=wav, sample_rate=48000, channel=mono, english_normalization=true
• Web preview: format=mp3, sample_rate=44100, bitrate=128000, channel=mono
• Podcast: format=mp3, sample_rate=44100, bitrate=192000–320000, channel=stereo if mixing music

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "text": "A clear example input",
  "voice_id": "your-voice-id",
  "speed": 1,
  "volume": 1,
  "pitch": 0,
  "emotion": "happy",
  "english_normalization": false,
  "sample_rate": 8000,
  "bitrate": 32000,
  "channel": "1",
  "format": "mp3",
  "language_boost": "Chinese"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/minimax/speech-2.6-hd" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters