Alias TTS for developers

Keep the client. Upgrade the workflow.

Point an ElevenLabs- or OpenAI-compatible client at Alias TTS. Keep the request shapes, authentication, and errors your application already understands—then let any call become an editable Studio project with automatic QA and a verifiable final.

curl -X POST \

- https://api.elevenlabs.io/v1/text-to-speech/rachel \

+ https://aliastts.example.com/v1/text-to-speech/rachel \

-H "xi-api-key: sk_..."

Compatibility

Two API dialects. One production pipeline.

Both interfaces converge on the same voices, cache, sentence-aware chunking, generation, cleanup, and assembly.

ElevenLabs-compatible

POST /v1/text-to-speech/{voice_id}, streaming, familiar headers, body fields, and error formats.

OpenAI-compatible

POST /v1/audio/speech with Bearer authentication and preset names that can map to your voices.

Built for long reads

Async jobs accept long scripts, while caching and controlled seams keep repeat work fast and finished audio clean.

Connection details, API keys, and voice IDs ready to copy into a client.

API → Studio

A successful request can remain editable.

Choose whether every API request, only a failed request, or no request leaves behind a project. When it does, the script is already split into chunks with its generated takes attached.

Repair without starting over

Edit or regenerate one chunk, compare takes, change the voice for a passage, and rebuild only when the project is ready.

Inspect before spending

Preview cleaned text, chunk boundaries, pronunciation changes, and estimated cost before committing to a render.

Automatic QA

Alias can compare each generated chunk to its script and flag or repair truncation, noise, and suspicious pauses.

Keys with boundaries

Per-key hourly budgets keep one integration from exhausting the generation capacity intended for the rest.

An API-created Studio project with editable chunks, saved takes, and inline QA state.

Verification

Ship the file with its proof.

Seal an approved build with the SHA-256 fingerprint of its exact audio bytes and a frozen record of the script, chunks, settings, and selected takes. Anyone can later check that the file is unchanged.

Portable receipt

Export the audio with human-readable provenance and a machine-readable manifest.

Public check

Drop a file on the verification page to confirm that it matches an approved final.

Local hashing

The browser computes the fingerprint locally; the audio itself is not uploaded for verification.

Public verification of a sealed, approved audio file.

Change the endpoint. Keep the integration.

And when a take needs human judgment, open the project in Studio.