MiniMax (Voice Clone and TTS)

Overview

MiniMax provides two speech routes in StoryFlow:

1. Voice Clone: clone timbre from a reference sample.
2. TTS: synthesize speech from text using preset or custom voice IDs.

1) Voice Clone

Inputs

• prompt (demo text, required)
• reference_audio (required)

Audio requirements (current config)

• MP3 / M4A / WAV
• roughly 10s to 5min
• up to 20MB

Common parameters

• voice_model (speech-2.5-hd-preview)
• accuracy
• need_noise_reduction
• need_volume_normalization

2) MiniMax TTS

Inputs

• prompt (required)

Common parameters

• voice_model: speech-2.6-turbo / speech-2.6-hd
• voice_id: preset system voice
• use_custom_voice + custom_voice_id
• emotion
• speed / vol / pitch

Tips

• Use clean single-speaker samples for cloning.
• Split long scripts for better pacing control.
• Keep voice settings consistent across scenes.