# Gladia Enterprise API

> Real-time speech-to-text API compatible with Gladia's Live V2 API: create a session with POST /v2/live, stream audio over its WebSocket, receive partial and final transcripts, translations, named entities, PII redaction, sentiment and end-of-turn events. Authentication: the x-gladia-key header. API addresses: https://eu1.api.gladia.io (Europe), https://us1.api.gladia.io (United States).

The complete documentation in one file: https://eu1.docs.gladia.io/llms-full.txt. OpenAPI document: https://eu1.docs.gladia.io/openapi.json.

## Get started

- [Welcome](https://eu1.docs.gladia.io/introduction.md): What the Gladia Enterprise API does, how a live session works, and where to go next.
- [Authentication](https://eu1.docs.gladia.io/introduction/authentication.md): API keys, their scopes, IP allowlists, and how to keep your key out of browsers and apps.
- [Models](https://eu1.docs.gladia.io/introduction/models.md): Compare solaria-3, solaria-1 and trantoria-1 by language, latency and features, and choose a latency profile.
- [Quickstart](https://eu1.docs.gladia.io/introduction/quickstart.md): Hand the integration to your coding agent, or get an API key, start a live session with the Gladia SDK and stream an audio file to get its transcript, in five minutes.
- [Regions](https://eu1.docs.gladia.io/introduction/regions.md): The API addresses of each region, and where your audio and transcripts are processed and stored.
- [SDKs](https://eu1.docs.gladia.io/introduction/sdks.md): Use the official Gladia SDKs for Python and JavaScript with the Gladia Enterprise API, and how to pass the options they do not know yet.
- [Playground](https://eu1.docs.gladia.io/playground.md): Transcribe your microphone live in the browser with your API key.

## Live transcription

- [Audio formats](https://eu1.docs.gladia.io/live/audio-formats.md): Encodings, sample rates, bit depths and channels the live API accepts, how to send the audio, and chunk sizes.
- [Callbacks](https://eu1.docs.gladia.io/live/callbacks.md): Receive a session's messages on your server by HTTP POST, verify their signature, and know how they are delivered and retried.
- [End of turn](https://eu1.docs.gladia.io/live/end-of-turn.md): Know when a speaker has finished their turn, from how they speak and not from silence alone, to make voice agents answer at the right moment.
- [Endpointing](https://eu1.docs.gladia.io/live/endpointing.md): Control when an utterance ends and its final transcript is sent, the maximum utterance length, and the number hold.
- [How a session works](https://eu1.docs.gladia.io/live/how-it-works.md): The life of a live session, from POST /v2/live to end_session, the messages in order, and how to choose which ones you receive.
- [Partial transcripts](https://eu1.docs.gladia.io/live/partial-transcripts.md): Get the words of an utterance while it is spoken, before its final transcript, for live captions and fast voice agents.
- [Recommended parameters](https://eu1.docs.gladia.io/live/recommended-parameters.md): Starting configurations for voice agents, call centers, meetings and live captions.
- [Reconnections](https://eu1.docs.gladia.io/live/reconnections.md): How a live session survives a network drop or a server restart without losing audio or transcripts, with the SDKs or your own client.

## Languages

- [Code switching](https://eu1.docs.gladia.io/languages/code-switching.md): Transcribe conversations that switch between languages, with the language detected among the ones you list.
- [Language detection](https://eu1.docs.gladia.io/languages/language-detection.md): Let the API detect the language spoken, or set it, and read the language of each utterance.
- [Supported languages](https://eu1.docs.gladia.io/languages/supported-languages.md): Every language the live API transcribes, by model, with its ISO 639-1 code.

## Audio intelligence

- [Custom metadata](https://eu1.docs.gladia.io/features/custom-metadata.md): Attach your own identifiers to a session, get them back with its results and callbacks, and find sessions by them; with the limits to know.
- [Custom spelling](https://eu1.docs.gladia.io/features/custom-spelling.md): Write your brand names, abbreviations and preferred spellings exactly as you want them in every transcript, step by step, with its limits.
- [Custom vocabulary](https://eu1.docs.gladia.io/features/custom-vocabulary.md): Make the transcription get your product names, people and jargon right, step by step with the SDKs, with the limits to know before you ship.
- [Named entity recognition](https://eu1.docs.gladia.io/features/named-entity-recognition.md): Find people, companies, places, dates, amounts, emails and phone numbers in each sentence as it is spoken; step by step, with the traps to avoid.
- [Numbers as digits](https://eu1.docs.gladia.io/features/numerals.md): Get amounts, quantities and numbers written with digits, ready to store or act on, step by step, with the limits to know.
- [PII redaction](https://eu1.docs.gladia.io/features/pii-redaction.md): Remove names, phone numbers, emails, card numbers and other personal data from transcripts before they reach your systems; step by step, with every trap to avoid.
- [Punctuation and casing](https://eu1.docs.gladia.io/features/punctuation.md): Readable transcripts with punctuation and capital letters, on by default; how to turn it off and what to expect from partials.
- [Sentiment analysis](https://eu1.docs.gladia.io/features/sentiment-analysis.md): Follow the sentiment and emotion of each sentence during a call, to alert a supervisor or score a conversation; step by step, with its limits.
- [Speaker diarization](https://eu1.docs.gladia.io/features/speaker-diarization.md): Know who said what when several people share one microphone, step by step, with its limits and when to use channels instead.
- [Translation](https://eu1.docs.gladia.io/features/translation.md): Translate what is said into other languages in real time, from the audio, with solaria-3; step by step, with every limit to know before you ship.

## Your data

- [Audio recordings](https://eu1.docs.gladia.io/data/audio-recordings.md): Download the audio of a live session, in FLAC, when your account records it.
- [History](https://eu1.docs.gladia.io/data/history.md): List, read and delete your past live sessions with the API, and filter them by date, status or metadata.
- [Data retention](https://eu1.docs.gladia.io/data/retention.md): How long sessions, transcripts and recordings are kept, and how to delete them earlier.

## Limits

- [Limits](https://eu1.docs.gladia.io/limits/concurrency.md): Concurrent sessions, session duration, timeouts and how to handle 429 and 503 answers.

## Migrations

- [From Gladia's cloud API](https://eu1.docs.gladia.io/migrations/from-gladia-cloud.md): Move an integration of Gladia's Live V2 API to the Gladia Enterprise API, and the differences to know.

## API reference

- [API reference](https://eu1.docs.gladia.io/api-reference.md): Base URLs, authentication, errors and every endpoint and message of the Gladia Enterprise live API.
- [Initiate a session](https://eu1.docs.gladia.io/api-reference/v2/live/init.md): Creates a live session with its options and returns the WebSocket URL to stream its audio to.
- [Live WebSocket](https://eu1.docs.gladia.io/api-reference/v2/live/websocket.md): The `url` of `POST /v2/live`. Send the audio as binary frames, or as `audio_chunk` messages; send `stop_recording` when done. The server sends the messages below, then closes the connection.
- [Get a session](https://eu1.docs.gladia.io/api-reference/v2/live/get.md): A session's status and options and, once it ended, its transcript.
- [List sessions](https://eu1.docs.gladia.io/api-reference/v2/live/list.md): Your account's sessions, newest first, one page at a time.
- [Download the audio](https://eu1.docs.gladia.io/api-reference/v2/live/get-audio.md): The session's recording, in FLAC, for accounts that record their audio. Available about 40 seconds after the session ends. A `Range` header returns part of it.
- [Delete a session](https://eu1.docs.gladia.io/api-reference/v2/live/delete.md): Deletes a session that has ended, its transcript and its recording.
- [Get the defaults](https://eu1.docs.gladia.io/api-reference/v2/live/defaults.md): The value each option gets when the init body omits it.
- [start_recording message](https://eu1.docs.gladia.io/api-reference/v2/live/message/start-recording.md): The session is ready: audio may be sent (`messages_config.receive_lifecycle_events`).
- [transcript message](https://eu1.docs.gladia.io/api-reference/v2/live/message/transcript.md): A partial (`receive_partial_transcripts`) or final (`receive_final_transcripts`)
transcript.
- [speech_start message](https://eu1.docs.gladia.io/api-reference/v2/live/message/speech-start.md): Speech started on a channel (`receive_speech_events`).
- [speech_end message](https://eu1.docs.gladia.io/api-reference/v2/live/message/speech-end.md): Speech ended on a channel (`receive_speech_events`).
- [named_entity_recognition message](https://eu1.docs.gladia.io/api-reference/v2/live/message/named-entity-recognition.md): The entities of a final, right after it (`receive_realtime_processing_events`).
- [pii message](https://eu1.docs.gladia.io/api-reference/v2/live/message/pii.md): The personal data redacted in a final, right after it (with `pii`; on the WebSocket
only).
- [sentiment_analysis message](https://eu1.docs.gladia.io/api-reference/v2/live/message/sentiment-analysis.md): The sentiment of a final, right after it (`receive_realtime_processing_events`).
- [translation message](https://eu1.docs.gladia.io/api-reference/v2/live/message/translation.md): One final translated into one target language (`receive_realtime_processing_events`).
- [end_of_turn message](https://eu1.docs.gladia.io/api-reference/v2/live/message/end-of-turn.md): The speaker finished their turn (`end_of_turn`): a voice agent may answer.
- [audio_chunk message](https://eu1.docs.gladia.io/api-reference/v2/live/message/audio-chunk.md): Audio transcribed: a client may drop it from its resend buffer
(`receive_acknowledgments`).
- [stop_recording message](https://eu1.docs.gladia.io/api-reference/v2/live/message/stop-recording.md): The answer to `stop_recording` (`receive_acknowledgments`).
- [end_recording message](https://eu1.docs.gladia.io/api-reference/v2/live/message/end-recording.md): No more audio is accepted (`receive_lifecycle_events`).
- [post_transcript message](https://eu1.docs.gladia.io/api-reference/v2/live/message/post-transcript.md): The whole session's transcript, after `end_recording`
(`receive_post_processing_events`).
- [post_final_transcript message](https://eu1.docs.gladia.io/api-reference/v2/live/message/post-final-transcript.md): The session's result, the same as `GET /v2/live/{id}`'s
(`receive_post_processing_events`).
- [end_session message](https://eu1.docs.gladia.io/api-reference/v2/live/message/end-session.md): The last message: the server closes the WebSocket after it (`receive_lifecycle_events`).
- [error message](https://eu1.docs.gladia.io/api-reference/v2/live/message/error.md): Something failed during the session (`receive_errors`).