{"id":49606336,"url":"https://github.com/jackdoe/speak","last_synced_at":"2026-05-04T13:08:09.833Z","repository":{"id":337060793,"uuid":"1152144220","full_name":"jackdoe/speak","owner":"jackdoe","description":"push to talk asr with whisper.cpp on metal","archived":false,"fork":false,"pushed_at":"2026-02-07T14:37:57.000Z","size":75,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"master","last_synced_at":"2026-02-07T23:17:12.485Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Swift","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jackdoe.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-02-07T12:36:57.000Z","updated_at":"2026-02-07T14:38:00.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/jackdoe/speak","commit_stats":null,"previous_names":["jackdoe/speak"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/jackdoe/speak","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jackdoe%2Fspeak","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jackdoe%2Fspeak/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jackdoe%2Fspeak/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jackdoe%2Fspeak/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jackdoe","download_url":"https://codeload.github.com/jackdoe/speak/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jackdoe%2Fspeak/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32608381,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-04T10:08:07.713Z","status":"ssl_error","status_checked_at":"2026-05-04T10:08:02.005Z","response_time":58,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-05-04T13:08:09.276Z","updated_at":"2026-05-04T13:08:09.827Z","avatar_url":"https://github.com/jackdoe.png","language":"Swift","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Speak\n\nPush-to-talk transcription for macOS. Hold a key, speak, release — your words appear wherever your cursor is.\n\nUses [whisper.cpp](https://github.com/ggml-org/whisper.cpp) with Metal GPU acceleration on Apple Silicon.\n\n## Features\n\n- **Push-to-talk** — F12 to transcribe, F11 to transcribe + send (hits Return)\n- **Menu bar app** — lives in the status bar, no dock icon\n- **Metal accelerated** — runs whisper.cpp on the GPU\n- **Voice activity detection** — filters silence, supports 10+ minute recordings with pauses\n- **Recording overlay** — floating indicator with VAD state, duration, and audio level\n- **Model management** — download models from HuggingFace directly in the app\n- **Configurable** — hotkeys, output mode (type/paste), VAD thresholds, all whisper.cpp parameters\n- **Instant start** — optional always-on microphone for zero-latency recording\n\n## Requirements\n\n- macOS 14+\n- Apple Silicon (M1/M2/M3/M4)\n- Xcode Command Line Tools\n- CMake (`brew install cmake`)\n\n## Build\n\n```bash\ngit clone --recursive https://github.com/user/speak.git\ncd speak\nmake app\n```\n\nThis builds whisper.cpp with Metal, compiles the Swift app, and creates `Speak.app`.\n\nTo install to `/Applications`:\n\n```bash\nmake install\n```\n\n## First Launch\n\nThe app guides you through setup:\n\n1. **Microphone permission** — for audio capture\n2. **Accessibility permission** — for global hotkeys and text output\n3. **Model download** — pick a whisper model (Large V3 Turbo Q5 recommended)\n4. **Preferences** — configure hotkeys and behavior\n\n## Usage\n\n- **F12** (hold) — record speech, (release) — transcribe and output text\n- **F11** (hold) — same as F12 but also presses Return after (for chat apps)\n- **Menu bar icon** — switch models, toggle mic warm, open settings\n\n## Make Targets\n\n| Command | Description |\n|---------|-------------|\n| `make whisper` | Build whisper.cpp with Metal |\n| `make build` | Build whisper.cpp + release binary |\n| `make debug` | Build debug binary |\n| `make run` | Build debug + run |\n| `make app` | Create Speak.app bundle |\n| `make install` | Copy Speak.app to /Applications |\n| `make uninstall` | Remove from /Applications |\n| `make clean` | Remove build artifacts |\n| `make models` | Download base.en + medium.en + large-v3 models |\n| `make models-large-turbo-q5` | Download large-v3-turbo-q5 (recommended) |\n\n## Project Structure\n\n```\nSources/Speak/\n├── App/              HotkeyManager, StatusBarController, SpeakApp\n├── Audio/            AudioEngine, RingBuffer, VoiceActivityDetector\n├── Whisper/          WhisperContext, ModelManager, ModelDownloader, WhisperSettings\n├── Pipeline/         TranscriptionPipeline\n├── UI/               SettingsView, RecordingOverlay, OnboardingView\n├── Utils/            TextOutput, PerformanceMonitor, LoginItemManager\n└── Benchmark/        BenchmarkRunner, BenchmarkView\n```\n\n## Models\n\nModels are stored in `~/Library/Application Support/Speak/models/`. Download via the app settings or CLI:\n\n| Model | Size | Speed (5s audio on M1) |\n|-------|------|----------------------|\n| tiny.en | 75 MB | ~0.15s |\n| base.en | 142 MB | ~0.3s |\n| small.en | 466 MB | ~0.8s |\n| large-v3-turbo-q5 | 547 MB | ~0.5s |\n| medium.en | 1.5 GB | ~2.5s |\n| large-v3 | 2.9 GB | ~5s |\n\n## Built by\n\nWritten by Claude Opus 4.6 in team mode — architecture, research, and implementation done by a coordinated team of specialist agents (ML engineer, audio engineer, UI engineer) working in parallel.\n\n## License\n\nwhisper.cpp is MIT licensed. This project wraps it with a native macOS interface.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjackdoe%2Fspeak","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjackdoe%2Fspeak","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjackdoe%2Fspeak/lists"}