TensorSpeak v0.3.8

What's new

• Redesigned settings into Try / Voices / Engine bottom tabs
• Chunking is a primary Engine control (with char budgets in the labels), not buried under Advanced
• Optional cloud voices and Advanced runtime (backend / threads) are collapsed by default
• System TTS gear opens the Engine tab

Assets

ONNX graphs are not bundled in the APK. Download once from this release (or from the in-app On-device models section) before using Micro / Nano:

• TensorSpeak-model-micro.zip — Inflect Micro v2 (~38 MB)
• TensorSpeak-model-nano.zip — Inflect Nano v2 (~16 MB)

APKs:

• TensorSpeak-0.3.8-arm64-v8a.apk — phones (arm64-v8a); Obtainium APK filter arm64
• TensorSpeak-0.3.8-x86_64.apk — x86_64 emulators / Chromebooks

Attribution and licensing:

• THIRD_PARTY_NOTICES.md
• docs/MODEL_ATTRIBUTION.md

TensorSpeak v0.3.7

Signed per-ABI APKs for TensorSpeak 0.3.7.

Controllable smart chunking: Advanced → Chunking profiles set both first-chunk and subsequent limits (Fast 64/160, Balanced 96/280, Continuous 280/560) while keeping the sentence/punctuation splitter.

ONNX graphs are not bundled in the APK. Download once from this release (or from the in-app On-device models section) before using Micro / Nano:

• TensorSpeak-model-micro.zip — Inflect Micro v2 (~38 MB)
• TensorSpeak-model-nano.zip — Inflect Nano v2 (~16 MB)

APKs:

• TensorSpeak-0.3.7-arm64-v8a.apk — phones (arm64-v8a); Obtainium APK filter arm64
• TensorSpeak-0.3.7-x86_64.apk — x86_64 emulators / Chromebooks

Attribution and licensing:

• THIRD_PARTY_NOTICES.md
• docs/MODEL_ATTRIBUTION.md