Text-to-Speech)¶

A single Google Cloud API key powers three optional backends:

Vision OCR — paid OCR engine (1,000 free / month)
Speech-to-Text v1 — paid STT (60 minutes / month free)
Text-to-Speech v1 — paid TTS (1 M characters / month free for WaveNet)

You only need to enable the APIs you actually use.

Get an API key¶

Create a Google Cloud project
Open the API library: https://console.cloud.google.com/apis/library
Enable any of:
Create an API key: click + Create Credentials → API key
Copy the key (looks like AIza...).

Restrict the key

On the API-key detail page, under API restrictions, restrict the key to just the APIs you've enabled. That way a leaked key can't rack up bills on services you didn't intend to use.

Configure in the app¶

In Settings → Service:

Paste into Google Cloud API key → Save

This single key is now available to all three Google services.

Enable each service¶

Vision OCR¶

In Settings → OCR → OCR method = Google Cloud OCR.

That's it — it'll use the same key from Service.

Speech-to-Text¶

In Settings → Subtitle → STT method = Google Cloud (for the Subtitle / Voice pages) or Settings → Live → STT method = Google Cloud (for the Live page).

In Settings → Subtitle → Google STT model, pick the recognition model:

Model	Best for
`latest_long` (default)	Long-form audio (interviews, lectures)
`latest_short`	Voice commands, short phrases
`phone_call`	Telephony audio (8 kHz)
`medical_dictation` / `medical_conversation`	Medical-domain audio

Text-to-Speech¶

In Settings → Voice → TTS method = Google Cloud TTS.

By default the server picks a voice based on language and gender — that's all most users need. Pinning a specific Google voice (e.g. en-US-Chirp3-HD-Charon, vi-VN-Wavenet-A) is supported by the engine but not yet exposed as a Settings field; it can be set by editing voice/google_tts_voice_name in settings.ini directly. Voice IDs are listed at https://cloud.google.com/text-to-speech/docs/voices.

Common errors¶

Error	Likely cause
`AUTH_ERROR`	Wrong / expired key. Re-paste in Settings → Service.
`API not enabled`	You haven't enabled the specific API (Vision / Speech / TTS) on this Cloud project.
`QUOTA_ERROR`	Free-tier limit reached for this API. Wait, or upgrade billing.
`INVALID_ARGUMENT_ERROR`	Voice name doesn't exist in the language you picked.

Cost guard¶

Warning

All three Google APIs are post-paid — once you exceed the free tier you start being billed without a stop. Set a budget alert on the Cloud project before doing high-volume work.