Use in apps
Any app that lets you choose an OpenAI-compatible speech-to-text server can send its recordings to Loquela Remote. It uses the same minutes and limits as the API.
The settings
- Server address (base URL):
https://api.getloquela.com/v1. Some apps add/v1themselves; for those, usehttps://api.getloquela.com. - Model:
loquela-remote-beta - API key: one of your keys from the account page, starting
lqr_ - Live or streaming text: off
The address and model are also on your account page, under API, with copy buttons. If an app asks for the full endpoint instead of a base address, use https://api.getloquela.com/v1/audio/transcriptions.
What works
- Sending a recording after you stop speaking, through
/v1/audio/transcriptions. - Checking a key:
/v1/modelslistsloquela-remote-betafor a valid key and returns 401 for a wrong one.
If an app reports an error, Troubleshooting covers the usual causes: a key copied incompletely, a base URL with /v1 missing or doubled, or a spoken language that is not on for your account.
Live text while you speak (OpenAI's realtime connection) is not available, so an app that only offers live transcription cannot use Loquela Remote. If the app sends a language, it has to be en, or one of the other languages when they are on for your account (language).
Loquela keyboard for Android
Setting up your own provider needs Loquela Full.
- Open Loquela's Settings, then Voice & Models.
- At the bottom, under Third-party provider, paste your
lqr_key into API key and tap Save key. Loquela recognizes it, shows Provider: Loquela Remote and fills in the server address and model itself. Test key should then say "The key works." - Tap Allow transcription with Loquela Remote, then Allow.
- Back in Settings, set Transcription location to API key (called Remote before Loquela 1.4).
In Loquela 1.6.2 and earlier, choose Custom server under Third-party provider first, enter the server address https://api.getloquela.com and the model loquela-remote-beta, tap Save, leave Server streams live text off, then save the key and allow it as above.
Each recording you start now goes from your phone to Loquela Remote, and the text comes back to the keyboard. Loquela Remote handles the audio as it does any API request (how your audio is handled), and the minutes show on your account page.
If you chose a Spoken language other than English in Voice & Models, Loquela Remote refuses the recordings unless other languages are on for your account. When it cannot be reached, the keyboard can transcribe on the phone instead: see If the provider cannot be reached in the same section.