[BUG] Voice dictation sends the dictated text twice when a character is typed through the Japanese IME before Enter (macOS desktop app, Code tab; reproduces every time)

Status Open
Reported on v2.1.296
Maintainer reply None cached
Activity 0 comments · opened Oct 13, 2026

Product: Claude desktop app (macOS), Code tab: the prompt box of a Claude Code session, using the app's own microphone (dictation) button
App version: Claude 2.31226.1 · Claude Code 2.1.296
OS: macOS 26.6.2 (build 25G83)
Keyboard / input source: macOS Japanese IME (com.apple.inputmethod.Japanese)
Dictation language: Japanese

What happened

With the Japanese IME as the active input source, typing anything by hand while dictation is active and then pressing Enter sends the dictated text twice. It reproduces on every attempt.

  • Dictate a message in Japanese with the app's microphone button, then press Enter without typing anything → the sent message is exactly the dictated text. Fine.
  • Dictate the same way, then type anything by hand while dictation is active (for example a 。 or 、, or any character entered through the Japanese IME) and press Enter → the sent message contains the dictated text twice.

Expected

The message is sent once, exactly as it appears in the box: the dictated text plus whatever was typed.

Reproduction

  1. Set the macOS input source to Japanese (Hiragana).
  2. In the Code tab prompt box, start dictation and say a sentence or two in Japanese.
  3. While dictation is still active, type a single character by hand (a 。 is enough).
  4. Press Enter.
  5. The sent message contains the dictation twice.

Notes

  • The trigger (type a character, then Enter) is the same as in #100792, which was observed once in English with a Latin keyboard. Here it reproduces every time with the Japanese IME. The IME commits text through composition events rather than plain key presses, which may be why the pending dictation segment is re-inserted instead of replaced; filing separately so the IME angle is not lost.
  • A separate Japanese dictation issue on the same build, a 。 inserted at the start of each new dictation segment, is #101656.

View original on GitHub ↗