Overview
Truebar is Vitasis’ end-to-end speech platform. It combines automatic speech recognition (ASR), natural-language-processing (NLP) stages, and neural speech synthesis (TTS) into configurable pipelines that you can run either as online streaming sessions or offline batch jobs. This page gives you the mental model you need before diving into the quick starts and API reference.
Why Truebar#
- Production ready pipelines — Compose ASR, punctuation, normalisation, translation, and TTS stages without hosting models yourself.
- Flexible deployment — Call the same pipelines over HTTP for offline tasks or WebSocket for low-latency streaming.
- Fine-grained access control — OAuth2 authentication plus per-stage roles keep your data and services scoped.
Core concepts#
Pipelines and stages#
Pipelines describe how audio and text flow through the platform. Each pipeline is made of stages, such as:
ASR– Streaming or batch speech-to-text.NLP_*– Text cleanup stages (punctuation, truecasing, text normalisation, inverse normalisation, accentuation, machine translation, autocorrect).TTS– Neural voice synthesis with optional SSML support.
Configure pipelines through JSON objects when opening a streaming session or submitting an offline job. Sensible defaults exist for common use cases, and discovery endpoints can recommend the best stage tags for your account.
Hosting environments#
Truebar ships in multiple environments (playground, production, bespoke deployments). Hostnames follow the pattern:
| Component | Playground | Production |
|---|---|---|
| Authentication | playground-auth.true-bar.si | auth.true-bar.si |
| API / Pipelines | playground-api.true-bar.si | api.true-bar.si |
| Web UI | playground.true-bar.si | editor.true-bar.si |
Swap the prefix to target your environment. Quick starts in this documentation default to the production hostnames.
Authentication & roles#
All endpoints require an OAuth2 access token issued by the authentication realm. Roles embedded in the token determine which stages or pipeline types you can use. For example:
PIPELINE_ONLINE_APIunlocks streaming WebSocket sessions.PIPELINE_OFFLINE_APIunlocks HTTP jobs.STAGE_TTS,STAGE_ASR,STAGE_NLP_TN, … control which stages you may include.
The Getting Started guide walks through obtaining and refreshing tokens.
Typical architecture#
[Microphone / Media Source] --> [Real-time pipeline] --> [Application logic] --> [Synthesis / Storage]Offline workflows skip the real-time leg and send media files straight to the HTTP pipeline API. Results (transcripts, diagnostics, audio) are stored and can be fetched via REST endpoints.
Next steps#
- Ready to try it out? Head to the Getting Started guide for environment setup and your first API calls.
- Building a real-time experience? Explore the Streaming STT quickstart and Streaming TTS quickstart.
- Working with batch jobs or archives? Review the Offline pipeline API and History API.
- Prefer examples? The API reference links to full pipeline and stage details.
Legacy APIs
Version 2.x documentation remains available under Truebar → Legacy in the navigation. New projects should target the 3.x pipeline APIs.