Skip to content

Releases: liznee/dsh-speech-input

Release list

v0.2.1

v0.2.1 Latest
Latest

Choose a tag to compare

View all tags
@liznee liznee released this 29 Aug 14:21
v0.2.1
9823c8a

Added / Changed

  • Auto-stop on silence: 5 s default (DEFAULT_SILENCE_TIMEOUT_MS), anchored to real speech (new transcript text or local mic RMS ? 0.04) - ambient noise and re-emitted results never extend it.
  • Listening pill rework: oval capsule with ring + shadow, end-cap curvature matching the circular buttons, thicker strokes, new five-bar volume stop icon, symmetric outlined cancel/stop, waveform baseline removed.
  • Published on npm.

Full history in CHANGELOG.md.

Assets 2
Loading

dsh-speech-input v0.1.0

dsh-speech-input v0.1.0

Choose a tag to compare

View all tags
@liznee liznee released this 22 Aug 03:23
v0.1.0
bd952e4

dsh-speech-input v0.1.0

First public release of the microphone dictation plugin for DeepSeek Harness Web.

Highlights

  • Dictates browser speech-recognition results into the Harness composer draft without auto-sending.
  • Uses the official composer slot and input APIs without modifying Harness DOM.
  • Preserves interim updates, manual suffix edits, cancellation, accessibility, and reduced-motion behavior.
  • Displays a live local RMS microphone meter and releases microphone resources on stop, cancel, error, or teardown.
  • Ships prebuilt lib/ artifacts, so Git installation does not need an allowBuilds grant.

Privacy note

The plugin does not store audio or send it to Harness or DeepSeek. Chrome and Edge Web Speech recognition commonly uses the browser vendor's online speech service, so recognition is not guaranteed to remain offline.

Install

dsh plugin --profile web add dsh-speech-input@0.1.0

GitHub installation:

dsh plugin --profile web add github:liznee/dsh-speech-input#v0.1.0
Loading