How rafin14/pockettts works
Pocket TTS is an offline Android reader and voice-cloning app. The main flow sends Reader text through chunked speech generation to cached audio, then foreground playback and reader controls; Android TTS clients have a separate entry point into the shared engine. Voice creation records or imports audio, optionally extracts it from video, trims and reviews the sample, optionally denoises it, then saves a named voice for synthesis. The README documents these capabilities; unsampled UI and extraction details are shown at subsystem-directory granularity. Edges reflect README statements or the supplied source index and excerpts.
Main entry points: VoiceRecordingDialog.kt Record & trim · ReaderPlaybackService.kt Reader playback · MainActivity.kt App UI · PocketEngine.kt Shared engine
Components 25
Reader & playback
- App UI
app/src/main/java/org/pockettts/android/engine/MainActivity.kt - Text chunking
app/src/main/java/org/pockettts/android/engine/ReaderTextChunker.kt - Reader playback
app/src/main/java/org/pockettts/android/engine/ReaderPlaybackService.kt - Audio timeline
app/src/main/java/org/pockettts/android/engine/AudioTimeline.kt - Reading highlight
app/src/main/java/org/pockettts/android/engine/ReadingHighlight.kt
Speech runtime
- TTS service
app/src/main/java/org/pockettts/android/engine/PocketTtsService.kt - Shared engine
app/src/main/java/org/pockettts/android/engine/PocketEngine.kt - Native Pocket TTS
app/src/main/java/org/pockettts/android/engine/NativePocketTts.kt - Pocket JNI
app/src/main/cpp/jni_bridge.cpp - Model & voice store
app/src/main/java/org/pockettts/android/engine/ModelPackRepository.kt - Bundled models
app/src/main/java/org/pockettts/android/engine/BundledPocketTts.kt
Voice creation
- Record & trim
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - Video audio import
app/src/main/java/org/pockettts/android/engine/VideoAudioExtractor.kt - Video import UI
app/src/main/java/org/pockettts/android/engine/VideoImportDialog.kt
Audio processing
- WAV inspection
app/src/main/java/org/pockettts/android/engine/WavSamples.kt - PCM WAV I/O
app/src/main/java/org/pockettts/android/engine/PcmWav.kt - DeepFilter denoiser
app/src/main/java/org/pockettts/android/engine/DeepFilterNet3Denoiser.kt - DeepFilter JNI & DSP
app/src/main/cpp/deepfilter/jni.cpp - DeepFilter DSP
app/src/main/cpp/deepfilter/deepfilter_dsp.cpp - KissFFT
app/src/main/cpp/deepfilter/kissfft/kiss_fft.c - Audio resampler
app/src/main/cpp/deepfilter/resampler.cpp - Sample preview
app/src/main/java/org/pockettts/android/engine/SamplePreview.kt
Other components
- Reader user
- Android TTS client
- Voice creator
Connections 25
- Reader user → App UI: edits text
README.md - App UI → Reader playback: starts playback
app/src/main/java/org/pockettts/android/engine/MainActivity.kt - Reader playback → Text chunking: chunks text
README.md - Reader playback → Shared engine: generates speech
app/src/main/java/org/pockettts/android/engine/ReaderPlaybackService.kt - Reader playback → Audio timeline: tracks audio
app/src/main/java/org/pockettts/android/engine/ReaderPlaybackService.kt - Reader playback → Reading highlight: updates sections
README.md - Android TTS client → TTS service: requests synthesis
README.md - TTS service → Shared engine: synthesizes text
app/src/main/java/org/pockettts/android/engine/PocketTtsService.kt - Shared engine → Native Pocket TTS: creates instance
app/src/main/java/org/pockettts/android/engine/PocketEngine.kt - Native Pocket TTS → Pocket JNI: calls JNI
- Shared engine → Model & voice store: loads settings
app/src/main/java/org/pockettts/android/engine/PocketEngine.kt - Bundled models → Model & voice store: prepares models
app/src/main/java/org/pockettts/android/engine/BundledPocketTts.kt - Voice creator → App UI: chooses workflow
app/src/main/java/org/pockettts/android/engine/MainActivity.kt - App UI → Record & trim: opens editor
app/src/main/java/org/pockettts/android/engine/MainActivity.kt - Video import UI → Video audio import: extracts audio
app/src/main/java/org/pockettts/android/engine/VideoImportDialog.kt - App UI → Video import UI: opens import
app/src/main/java/org/pockettts/android/engine/MainActivity.kt - Record & trim → WAV inspection: inspects sample
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - Record & trim → PCM WAV I/O: reads and writes
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - Record & trim → DeepFilter denoiser: enhances sample
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - Record & trim → Sample preview: previews audio
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - Record & trim → Model & voice store: saves voice
app/src/main/java/org/pockettts/android/engine/VoiceRecordingDialog.kt - DeepFilter denoiser → DeepFilter JNI & DSP: runs enhancement
- DeepFilter JNI & DSP → DeepFilter DSP: uses DSP
app/src/main/cpp/deepfilter/jni.cpp - DeepFilter DSP → KissFFT: uses transforms
app/src/main/cpp/deepfilter/deepfilter_dsp.cpp - DeepFilter JNI & DSP → Audio resampler: resamples audio