AssemblyAI ships TTS-1 API at $15 per 1M characters
Original title: AssemblyAI Ships AssemblyAI-TTS-1: A $15 per 1M Character Text-to-Speech API for Voice Agents
The hosted model runs only in US and EU clouds, with six languages, instant cloning and no open weights.
AssemblyAI made AssemblyAI-TTS-1 generally available as a streaming text-to-speech API for live voice agents, priced at $15 per 1 million characters with unlimited pay-as-you-go concurrency. It is cloud-only in us-east-1 and eu-north-1, supports English, Spanish, French, German, Italian and Portuguese, and offers more than 4,000 voices plus instant cloning that requires speaker permission. AssemblyAI’s October 28 tests claim 93.7% entity accuracy and 47 ms median time to first byte; Cekura’s independent run scored 81.5% reading accuracy and 159 ms median streaming latency. SSML, speed control, pronunciation dictionaries and emotion tags are not available yet.
Unconfirmed: Latency and accuracy figures that AssemblyAI reports are self-reported and differ from Cekura’s independent benchmark.
- HTTP, WebSocket and an OpenAI-compatible route; output is PCM, mu-law or A-law from 8 kHz to 48 kHz.
- Self-reported word accuracy is 99.0% on Seed-TTS and LibriSpeech; predicted MOS is 3.85.
- Cekura ranked streaming latency 8th of 19 timed models; Inworld TTS-2 Flash posted 45 ms at the same list price.
- A clone example used 24 seconds of audio; LiveKit and Pipecat reference agents were published.
- Vendors
- AssemblyAI
- AI models
- AssemblyAI-TTS-1
From the original article
8 min read at marktechpost.com
AssemblyAI has released AssemblyAI-TTS-1, its first text-to-speech model for voice agents. It costs $15 per 1M characters, offers unlimited concurrency without a contract, and supports 6 languages at launch. The post AssemblyAI Ships AssemblyAI-TTS-1: A $15 per 1M Character Text-to-Speech API for Voice Agents appeared first on MarkTechPost.
By Michal Sutter · marktechpost.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.