title: “Voca 1.4.2: diagnostic uploads and lighter speech tracking”
tags:

• voca
• android
• text-to-speech
• accessibility
• offline

Voca 1.4.2: diagnostic uploads and lighter speech tracking

Voca 1.4.2 addresses a failed diagnostic-upload path and reduces the work done while speech is playing. It is a patch release for the Android listen-later reader.

Some opt-in error reports failed with HTTP 415 because the previous Blossom host accepted a preflight request but rejected the encrypted binary upload. The app now uses a host that passed a signed binary upload and a hash-checked download. Sending a report still requires your approval in the app. This check does not prove that every private inbox delivery succeeds, so manual log export remains available.

Speech tracking does less work at sentence boundaries and during estimated word highlighting. Voca no longer stops an idle rendered-audio player at each live-speech handover, skips trace formatting when diagnostics are off, and schedules the next fallback word update near its expected time. On a Pixel 6a, the production build kept playing with its wake lock held at two screen-off checkpoints. That is a background lifecycle check, not an audible test of every speech engine.

This update cannot repair pauses produced inside a third-party TTS voice. If you still hear stutter, an app log and the engine and voice name will help isolate it.

The Voca 1.4.2 APK is production-signed; its SHA-256 is 4e52696531ed9a37fa1b7f55e2e3aafd995931b832646d58bf545042f4e6d1c2. You can also update through Zapstore. The source code is available under MIT.