Palabra.ai offers real-time speech translation, TTS and STT in 60+ languages for calls, events and APIs. Plans from $60/month; pay-as-you-go API.
Researched and fact-checked by Bartu Cavusoglu · Updated October 26, 2026
Palabra.ai is a real-time speech AI platform built around three models: speech-to-speech translation, text-to-speech and speech-to-text. The company describes its goal as removing language barriers so people and teams can communicate naturally in whatever language they speak.
It is sold in two forms. One is a set of ready-made tools for translating video calls, webinars, in-person events and live streams. The other is an API platform for developers who want to embed translation or voice features in their own products.
Palabra.ai is based in London. Its site reports 500,000+ translated minutes, 60+ languages, under one second of latency and 99% accuracy, and lists customers such as Walcon Virtual and GIS Group.
For calls, a user pastes a Zoom, Google Meet or Microsoft Teams link, selects the languages and voice, and clicks Join call. A bot joins the meeting and translates what each speaker says, delivering translated audio or captions to participants. No app is required for the host, and webinar viewers open a link, pick a language and listen or read.
For developers, one streaming API handles the pipeline of speech recognition, translation and TTS over WebRTC or WebSocket. The site shows an SDK example in which microphone audio in English is sent to speakers in Spanish.
Palabra.ai says the translation model is its own proprietary language model, which it says gives control over speed and quality instead of chaining third-party services.
On the translation side, Palabra.ai supports two-way conversation mode and one-to-many presentation mode, live captions and transcripts, automatic language detection and custom glossaries so industry terms stay consistent across languages. It also lists noise suppression and music isolation.
Voice handling is a focus. The platform clones the speaker's voice automatically, offers a library of voices modeled after professional interpreters, and uses diarization to give each speaker a distinct voice. Emotion transfer is listed as coming soon.
For developers, the standalone TTS model is advertised at 35 ms to first audio, and the STT model offers streaming transcription with timestamps and turn detection. Regional server deployment is available for lower latency.
Palabra.ai is a real-time speech AI platform that provides live speech-to-speech translation, text-to-speech and speech-to-text. It is available as ready-made tools for calls, events and streams, and as an API for developers.
Subscriptions are Starter at $60 per month, Pro at $150 and Team at $500, with lower rates of $45, $113 and $375 on yearly billing. API usage is pay as you go, from $0.002 per minute for STT to $0.04 per minute for speech-to-speech. Business, event and broadcaster packages are quoted by sales.
Yes. The site offers a free trial for translating calls with easy cancellation. Developers also get $50 in free credits when they sign up for the API.
It works with Zoom, Google Meet and Microsoft Teams for calls and webinars. For streams it integrates with OBS, vMix, YouTube, Vimeo, Castr and Cloudflare through SRT and RTMP.
The site lists 60+ languages for translation, and says additional languages can be added on request. All plans include access to the full language set.
The company states that conversations are encrypted and that audio and text are never stored or used for training. It describes this as zero data retention, and publishes a Trust Centre covering security, privacy and compliance.
Yes. Attendees scan a QR code, choose a language and listen or read live captions on their own devices. Event pricing is custom and quoted by sales.
Yes. The API covers speech recognition, translation and TTS over WebRTC or WebSocket, with SDKs, documentation and browser playgrounds. Features include language detection, diarization, voice cloning and custom glossaries.