Camb AI Releases MARS5 TTS: A Novel Open Source Text to Speech Model for Insane Prosody
In what ways does MARS5 TTS differ from GPT and Gemini models?
MARS5 TTS differs from GPT and Gemini models as it specializes in text-to-speech synthesis, featuring a unique two-stage AR-NAR architecture. While GPT and Gemini focus on text generation and understanding, MARS5 excels in producing high-quality, controllable speech output. Its use of DDPM in the NAR stage and prosodic control through text formatting sets it apart in speech synthesis.
What are the two inference modes available in MARS5 TTS?
MARS5 TTS offers two inference modes: a fast "shallow clone" that doesn't require the reference audio's transcript, and a slower but higher-quality "deep clone" that utilizes the prompt transcript. The shallow clone provides a quicker synthesis, while the deep clone offers enhanced quality and cloning accuracy at the cost of longer processing time.
How long should the reference audio be for optimal results in MARS5 TTS?
The reference audio for MARS5 TTS should be between 2-12 seconds long, with 6-second samples yielding optimal results5. The system supports two inference modes: a fast "shallow clone" that doesn't require the reference audio's transcript, and a slower but higher-quality "deep clone" that utilizes the prompt transcript.