# Is AssemblyAI-TTS-1 SOTA?

Pending benchmarks. Announced Oct 27, 2026; no independent results yet.

As of Nov 3, 2026, AssemblyAI-TTS-1 is pending benchmarks on isitsota: announced Oct 27, 2026, with no independent results on the boards we track yet.

AssemblyAI's first text-to-speech model, aimed at voice agents: streaming audio over HTTP or WebSocket, more than 4,000 voices and instant voice cloning, in English, Spanish, French, German, Italian and Portuguese. Generally available on AssemblyAI's API at $15 per 1M characters with no concurrency limit or annual contract. No weights.

Canonical: https://isitsota.com/m/assemblyai-tts-1
Last updated: 2026-10-28
Tracking since: 2026-10-28

## Announced facts

| Fact | Value |
| --- | --- |
| Lab | AssemblyAI |
| Announced | Oct 27, 2026 |
| API | Live (`assemblyai-tts-1`) · assemblyai-tts-1: HTTP POST /v1/speech and WebSocket /v1/ws on tts.us.assemblyai.com, an OpenAI-compatible route, and an EU host |
| Open weights | Closed · None |
| Task | Text-to-speech for voice agents; operates on continuous audio representations rather than discrete audio tokens |
| Languages | English, Spanish, French, German, Italian and Portuguese, with several accents; more languages promised |
| Voices | 4,000+ library voices plus instant voice cloning; no fine-tuned ("Pro") cloning yet; pronunciation dictionaries and emotion tags coming |
| Pricing | $15 per 1M characters; unlimited concurrency, no annual contract |
| Training data | Millions of hours of audio, including real conversational data, per AssemblyAI |

Tags marked unsure are not stated on the lab's own pages.

## What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

- [x] Publicly announced with a source
- [x] Usable by the public (API or weights)
- [ ] Listed on an independent board we track
- [ ] Enough results for a rank in a category
- [ ] Holds across two snapshots (crowns only)

## What the lab claims

Self-reported. Shown for context; not used in any verdict.

- Lab claim: 47 ms median time to first byte in AssemblyAI's head-to-head with five other TTS models (next fastest 119 ms), and 90 ms time to first audio in production.
- Lab claim: 99.0% word accuracy, 93.7% entity accuracy, 3.85/5 average on preference predictors such as UTMOSv2, 85.5 on TTSDS prosody and 0.64 cloned-voice speaker similarity.

## Sources

- LAB: [AssemblyAI blog: AssemblyAI-TTS-1 text-to-speech API](https://www.assemblyai.com/blog/text-to-speech-api) (2026-10-27)
- LAB: [AssemblyAI docs: text-to-speech voices](https://www.assemblyai.com/docs/tts/voices)
- LAB: [AssemblyAI on X: AssemblyAI-TTS-1](https://x.com/AssemblyAI/status/2115470615583056042) (2026-10-28)

Tracking continues until an independent board lists it.

## Related

- Lab: [AssemblyAI](https://isitsota.com/labs/assemblyai.md)
- All pending releases: https://isitsota.com/releases.md
- JSON: https://api.isitsota.com/v1/models/assemblyai-tts-1