MOSS-TTS has been released
A new open-source voice model just outperformed OpenAI's offering in key tests.
Deep Dive
A new open-source text-to-speech model called MOSS-TTS has been released, and according to community-submitted benchmark results on Seed TTS Eval, it has outperformed OpenAI's TTS model. The leaked evaluation scores suggest superior performance in key audio quality metrics. This development could significantly disrupt the current landscape of voice synthesis tools, which has been dominated by proprietary models from major AI labs.
Why It Matters
This could democratize high-quality voice synthesis and pressure closed-source leaders like OpenAI to lower prices.