Try the new DeepSeek V4 Flash today. Frontier intelligence at a fraction of the cost. Here

Text to speechPartner Model

Maya 2

A real-time Hindi text-to-speech model for natural, low-latency conversational voice experiences.

Model details

Developed by
Maya Research
Use case
text to speech
Size
2B

Maya 2 is Maya Research’s flagship real-time text-to-speech model, built natively for each supported language to preserve authentic accents, rhythm, and pronunciation. Designed for conversational voice applications, it delivers natural-sounding speech with low latency, making it well suited for responsive voice agents and other interactive experiences.

Maya 2 is available through Maya’s first-party APIs in both the United States and India, backed by Baseten infrastructure. Choose the regional API closest to your users for lower-latency text-to-speech. Each request supports up to 130 words of input and can generate up to 100 seconds of audio. Try Maya 2 through the first party API.

For dedicated infrastructure, you can:

Input

1curl -X POST https://$MAYA_ENDPOINT/v1/tts \
2  -H "Authorization: Bearer $MAYA_API_KEY" \
3  -H "content-type: application/json" \
4  -d '{"voice":"Ananya",
5       "text":"नमस्ते! आपका ऑर्डर कल पहुँच जाएगा।",
6       "language":"hi"}' \
7  --output out.pcm