feat: add MiniMax Cloud TTS as alternative speech provider

[octo-patch]

Summary

Adds MiniMax Cloud TTS as a new tab in the Gradio UI, giving users a GPU-free alternative for text-to-speech synthesis via the MiniMax T2A V2 API.

What is included

• minimax_tts.py Standalone MiniMax TTS client with three output modes:
  • synthesize() raw MP3 bytes
  • synthesize_to_file() WAV or MP3 file
  • synthesize_to_tensor() PyTorch tensor (compatible with Chatterbox pipeline)
• Cloud TTS (MiniMax) Gradio tab UI with API key input, text box, file upload, model/voice selection, speed control, and audio preview
• 12 built-in voices English_Graceful_Lady, Deep_Voice_Man, Friendly_Person, sweet_girl, cute_boy, and more
• 2 model tiers speech-2.8-hd (high quality) and speech-2.8-turbo (fast)
• 25 unit tests + 5 integration tests covering API calls, error handling, file output, and tensor conversion
• README updated with Cloud TTS documentation, setup instructions, and programmatic usage examples

Why

Not everyone has a GPU powerful enough for local Chatterbox synthesis. MiniMax Cloud TTS provides a high-quality cloud alternative that requires only an API key no VRAM needed.

Files changed (6 files, ~784 additions)

File	Change
minimax_tts.py	New: MiniMax TTS client module
Chatter.py	Add Cloud TTS tab + helper functions
requirements.txt	Add requests dependency
README.md	Add Cloud TTS docs + TOC entry
tests/test_minimax_tts.py	25 unit tests
tests/test_minimax_tts_integration.py	5 integration tests

Test plan

• 25 unit tests pass (pytest tests/test_minimax_tts.py)
• 5 integration tests pass with real API key
• Manual: launch python Chatter.py, open Cloud TTS tab, enter API key, generate speech
• Manual: verify MP3 and WAV export both work
• Manual: verify different voices and models produce different audio