Assess your English speech and practice targeted exercises to improve it — entirely in your browser.
SpeechImprover is a free, open-source, client-side web app. Record a reading passage, an improv prompt, or your own text; it analyzes ten speaking attributes locally, scores you against desirable / undesirable reference patterns, and tracks your progress over time. No account and no backend of its own — nothing leaves your device unless you turn on the optional Google Drive sync, which syncs into your own Drive (audio excluded).
- 🔗 Live app: https://adervec.github.io/SpeechImprover/
- 📂 The app lives in
speechimprover/— see its README to run it locally. - 🏗️ Architecture & analysis design:
design.md
- This is not professional advice. SpeechImprover is an experimental, educational tool built by a software developer — not a doctor, speech-language pathologist, therapist, voice coach, or lawyer. Its analysis is heuristic and not clinically validated. For any speech, voice, language, or hearing concern, consult a qualified professional. → Full disclaimer
- Privacy. Your recordings, analysis, and history stay in your browser; the project has no servers of its own and collects nothing. Two opt-in caveats: live transcription uses your browser's Web Speech API (which in Chrome/Edge may send audio to Google for processing), and the optional Google Drive sync writes your sessions/profile/settings — never audio — into a private folder in your own Drive. → Privacy notes
- 🎙️ Record & analyze reading passages, free / improv speech, articulation drills, or your own text.
- 📊 Ten attributes scored 0–100 (pace, clarity, filler economy, richness, resonant depth, expressiveness, non-nasal tone, no-uptalk, projection, breath & pausing).
- 🎯 Target-vs-anti-pattern "alignment" scoring against reference profiles.
- 🎭 Free-speaking with a chosen target tone/emotion; a review step (submit / restart / abandon) with transcript before analysis.
- 📁 Projects — group recordings toward a goal, e.g. read a whole book passage by passage with tracked position.
- 🧠 Targeted practice programs generated from your weakest attributes.
- 📈 Trends, a GitHub-style activity heatmap, 🕘 full recordings/history with A/B compare, repeat & per-clip download.
- 🤝 AI-coach export (a machine-readable progress summary), 🔁 JSON import/export, and 🎨 themes.
- ☁️ Optional Google Drive sync (into your own Drive; audio excluded) · 📱 installable PWA (works offline).
cd speechimprover
npm install
npm run dev # dev server
npm run build # production build to dist/Full instructions and browser notes are in speechimprover/README.md.
React 19 + Vite, a pure client-side SPA. Continuous integration (lint + build) and deployment to
GitHub Pages run via GitHub Actions (see .github/workflows/).
- Reading passages are excerpts from public-domain works via Project Gutenberg (Austen, Dickens, Tolstoy, and others).
- The "Baseline guide" of speaking principles was generated for this project.
- Built with React and Vite (both MIT-licensed); all dependencies are permissive-licensed (MIT/Apache-2.0/ISC/BSD).
MIT © 2026 Adam Eryavec. Provided "as is", without warranty of any kind — see the disclaimer.