Text-to-Speech (TTS)

Use MiniMax TTS to generate speech from text (text-to-speech). Supports Chinese (Mandarin), English, and Japanese preset or custom voices. Output MP3 is automatically imported as AudioClip, saved to Assets/TJGenerators/History/.

Suitable for character voiceover, narration, notification sounds, dialogue lines, etc. For background music, use the BGM from Audio Generation; for sound effects, use SFX.

Getting Started

Start from GUI

Menu: AI > Generate > Generate Audio, in the opened TJGenerators Audio window, switch the "Model" to MiniMax Text-to-Speech to use TTS: enter the text to synthesize, select a voice character, and click generate.

Start from CLI

Use the generate_tts tool to generate speech from text. After generation, the speech is automatically imported as an AudioClip and ready to use in your scene.

---

Models

MiniMax Text-to-Speech (Only Model)

• generator_id: minimax-tts
• Use Cases: Text-to-speech, character voiceover, narration
• Output: MP3, automatically imported as AudioClip
• Key Parameters:

Parameter	Type	Default	Description
prompt	string	Required	Text to synthesize
voice_id	string	Chinese (Mandarin)_Gentleman	Voice ID (see table below, supports custom)
output_path	string	Auto-generated	Asset save path (.mp3 auto-appended)
play_on_awake	bool	false	Whether AudioSource auto-plays on entering Play Mode

Preset Voices (voice_id)

value	Description
Chinese (Mandarin)_Gentleman	Chinese Male - Gentleman (default)
Chinese (Mandarin)_Humorous_Elder	Chinese Male - Humorous Elder
Chinese (Mandarin)_Cute_Spirit	Chinese Female - Cute Spirit
Chinese (Mandarin)_Warm_Bestie	Chinese Female - Warm Bestie
English_WiseScholar	English Male - Wise Scholar
English_captivating_female1	English Female - Captivating
Japanese_LoyalKnight	Japanese Male - Loyal Knight
moss_audio_f0c5494c-7c25-11f0-8d70-a2abf1fbea61	Japanese Female

voice_id supports custom values (allowCustom), you can enter other voice IDs provided by MiniMax.

---

Optimization

Text Optimization

• Use the language matching the voice: Chinese voice for Chinese text, English voice for English text
• Add appropriate punctuation: Commas and periods help control pauses and intonation
• Segmented synthesis: Long dialogue lines can be split into multiple sentences for separate generation, making it easier to trigger them in the engine as needed

Voice Selection

• Character dialogue: Choose voices matching character gender/personality (Gentleman/Humorous Elder/Cute Spirit, etc.)
• Narration/commentary: Prefer steady male or female voices
• Multi-language projects: Use voices in the corresponding language for each part

AudioSource Configuration

• One-time lines/notification sounds: play_on_awake: false, triggered by scripts or events
• Narration that plays on entering scene: play_on_awake: true

---

Notes

• ⚠️ Entry point: GUI via AI > Generate > Generate Audio window, switch model to "MiniMax Text-to-Speech"; CLI via generate_tts tool
• ⚠️ prompt is required: The text to synthesize cannot be empty
• ⚠️ Language and voice must match: Chinese text with Chinese voice, otherwise pronunciation may be abnormal
• ⚠️ Generation takes 10–30 seconds
• ⚠️ Output is AudioClip (MP3): When placing in scene, bind as sound effect (AudioClip SFX) to AudioSource
• ⚠️ Output path: Default Assets/TJGenerators/History/
• ⚠️ Domain Reload: Do not write .cs files to disk during generation, use execute_csharp_script instead