Jose Will

Project 06 · Language learning / Daily practice

Daily Japanese & English Shadowing

A Tokyo-dated archive with one complete 30-minute C1+ Japanese speech plus 30 minutes of C1+ English speech and small talk.

DAILY SHADOWING · 60 MIN

Practise long-form speech for one hour.

A complete C1+ Japanese speech for 30 minutes, followed by 30 minutes of English C1+ speech and small talk, refreshed and archived on Tokyo time.

Speed

Preparing the long-form lesson…

These are original C1+ practice scripts and do not reproduce the linked source text.

Problem

How can the site replace elementary one-liners and nominal duration labels with genuine C1+ long-form material and dependable Japanese and English playback?

Role

Lesson design, source policy, scheduled generation, speech playback, archive UI, and implementation

Process

  1. Use one complete C1+ Japanese speech and C1+ English speech plus small talk, with tested minimum character and word counts
  2. Lazy-load Piper Plus only when playback is requested, synthesize natural Japanese and English locally, cache the model, and reserve system speech as a last fallback
  3. Archive lessons by Tokyo date in Durable Object v2, isolate obsolete short-form records, and keep past sessions selectable

Outcome

The working version enforces at least 4,000 Japanese characters and 3,000 English words, with natural neural speech, speed controls, paragraph playback, and dated archives.

Reflection

Practice time must be supported by real content volume and a protocol, not a label; speech playback must also account for asynchronous voices and browser length limits.