MiniMax Speech HD AI Voice Generation

Affordable text-to-speech and voice cloning in 29 languages.

MiniMax Speech HD is MiniMax's high-definition text-to-speech model, released in February 2025. It supports 29 languages and voice cloning from a 30-second sample, running somewhat cheaper than ElevenLabs v3 with comparable quality for most use cases. On Oakgen, a 1-minute narration (up to 1,000 characters) costs about 20 credits (~$0.08).

1. 29 languages with preserved timbre across languages
2. 30-second voice cloning from a clean sample
3. 44.1 kHz MP3/WAV output
4. 2–5 second latency per request
5. Cheaper per character than ElevenLabs v3

1. Pick a stock voice or clone your own

   Browse 150+ stock voices by language, age, and use case. For a cloned voice, upload a 30-second clean sample.

2. Paste your script

   Up to 50,000 characters per request — about 45 minutes of audio. Use punctuation for natural pauses.

3. Generate

   Typical 1-minute script finishes in 3–5 seconds. Download MP3 or WAV.

curl -X POST https://api.oakgen.ai/v1/generate/speech \
  -H "Authorization: Bearer $OAKGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax-speech-hd",
    "voice": "female_alloy",
    "text": "Welcome to Oakgen."
  }'

vs. elevenlabs-v3 — cost

MiniMax Speech HD costs somewhat less than ElevenLabs v3 with comparable quality for most narration. ElevenLabs v3 wins on emotional nuance, breath, and dramatic dialogue — pick it for audiobooks or cinema.

Licensed through MiniMax's commercial terms.

Permitted on all paid Oakgen plans.