A practice studio, not another lesson tree
Everyone Can Use English is a Chinese-language project built around a book, a 1,000-hour curriculum, and an application called Enjoy. Instead of presenting a fixed ladder of beginner lessons, Enjoy lets learners bring audio, video, web pages, and ebooks into a workspace for close listening, recording, and repetition.
That differs from a streak-driven language app. The core audio flow splits material into sentences, loops a line or phrase, records your imitation, and compares your pitch contour with the source. It can transcribe audio, create speech for text, translate passages, and turn ebook paragraphs into speaking exercises. The browser extension adds practice on YouTube and Netflix.
The catch is audience. The main README, curriculum, installation pages, and most issue discussion are in Chinese. A tiny English developer README for the Electron app contains only yarn install and yarn start. English speakers can inspect the code, yet there is no equivalent English user guide.
Shadowing is where Enjoy earns its place
Enjoy is strongest when you have a recording worth copying. On first import, it analyzes the waveform and pitch and uses speech recognition to produce subtitles. It then breaks the audio into sentences for repeated playback. You can select words or phrases, record yourself, and listen to the source and your attempt together. This is more focused than repeatedly scrubbing a podcast player.
The pitch display gives a visible clue about rhythm and intonation, but the project is honest about what formal assessment can judge. The documented pronunciation assessment uses Microsoft Azure, focuses more on whether individual words sound correct, and cannot reliably judge pitch movement. It is also a paid Enjoy feature even when you bring your own OpenAI key for other AI tools. Open source does not make every cloud-backed operation free.
The media support broadens the value. Local ebooks can be EPUB, text, or Markdown, with per-paragraph translation, speech, and follow-along practice. Videos use roughly the same flow, although the documentation caps imported video at 100 MB. YouTube importing requires a network that can reach YouTube.
Hosted convenience hides service dependencies
The fastest trial is enjoy.bot. Desktop downloads cover Windows 10 or newer, Apple silicon and Intel Macs, and x64 Linux. Local development is less polished. The monorepo expects Node 20 or newer and Yarn 4.6, while the Electron workspace brings in speech services, FFmpeg, SQLite, dictionaries, Playwright, and a large front-end stack. Its start task downloads dictionaries. There is no clear architecture or end-to-end self-hosting guide in the app README.
Speech-to-text can run through the bundled whisper.cpp integration, but the FAQ warns that it fails on some lower-specification machines and older operating systems. Cloud speech services are the fallback. AI features can use Enjoy's paid service or a configured provider key. Pronunciation scoring still requires an Enjoy balance. This is not a completely offline tutor.
The FAQ says most app data lives locally in an EnjoyLibrary folder. That is good for control, but Enjoy does not provide cloud synchronization. Its multi-device workaround is to sync that folder with a drive service, wait before switching computers, and avoid simultaneous use because conflicts can occur.
Current faults matter to the learning loop
The open queue shows present friction. An August 12, 2026 report says pronunciation evaluation recognized only part of a full recording. Other recent reports describe web imports without automatic sentence splitting, subtitle generation failures, audio that will not play, and assessment stuck after a browser content-security-policy error. One user says automatic cuts can clip the start or end of sentences. These faults affect the listen, record, assess cycle that makes Enjoy distinctive.
Network sensitivity is another recurring theme. The official FAQ recommends alternate API settings or a proxy when login and assessment calls time out, and explains that YouTube may require proxy configuration. July 2026 reports describe playback failures despite apparently working connections. Test your own material before investing time in a library.
Documentation is caught between product generations. The repository says the new desktop version will wrap and enhance the web version, while downloadable v0.7.9 dates to March 2025. A documentation page warns that development moves quickly and descriptions may lag. A recent request asks for basic guidance covering web and desktop differences, importing, TTS configuration, and feature use. The older guides do not map every current screen.
Healthy project, qualified recommendation
The code was last pushed on June 29, 2026, and issues were receiving updates in August. Its 124 open items include issues and pull requests, so that number is a workload signal rather than a bug count. The old release tag does not mean abandonment: current pushes and issue activity show continued work.
Enjoy is worth trying if you speak Chinese, learn best by shadowing, and want to turn personal media into exercises. It offers a better practice loop than a normal player and more control than a packaged course. Keep backups, test recognition with your accent and browser, and budget for assessment if that feedback is central to your routine.