
1. Technology overview
Kalpa Labs is a San Francisco company in the Y Combinator Fall 2025 batch. It trains speech models with the stated aim of one generalist audio model, and sells API access to its first model.
The kalpa-tts-1 API converts text to speech in eleven voices in English and Hindi, including English words inside Hindi sentences. The site states about 300 ms to first audio with streaming output over WebSocket and WAV or PCM formats, a realtime demo, and a public audio realism benchmark. A waitlist exists for the future generalist model.
2. Highlighted features & specifications
- Text to speech API: converts text to audio via REST.
- Streaming: WebSocket output with about 300 ms to first audio.
- Voices: eleven voices in English and Hindi.
- Code-switching: handles English words inside Hindi text.
- Output formats: WAV and 24 kHz 16-bit PCM.
- Benchmark: public AudioRealismBench comparison page.
3. Potential application
A contact center can prototype an English voice bot for order updates. A product team can add narration to an app in English or Hindi.
4. Benefits
Low latency suits voice agents. Hindi and code-switching support covers Indian-language markets. A free allowance of 1,000 minutes for 3 months lets teams test before a contract.
5. Market trends & opportunities
Voice agents for customer support are a growing use of speech models, and demand for non-English voices is rising. Vietnamese contact centers could use such APIs once Vietnamese voices are supported, which the site does not state.
6. Other information
Part of Y Combinator Fall 2025. Also publishes research reports and a leaderboard. Current language support is English and Hindi only.
One platform connecting startups, corporations, research institutions, regions and investors











Playbook & Handbook · February 6, 2026
Winning Accelerator Playbook 2026 by BambuUP
RETAIL E-COMMERCE: AI AND REAL-TIME DATA ARE TRANSFORMING HOW RETAIL BUSINESSES OPERATE








































