[FEATURE] Add auto-send after silence timeout for desktop app voice dictation

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

In the Claude Desktop app, voice dictation currently works like this: click the microphone to start recording, speak, then click again to stop. After stopping, the transcribed text sits in the input box and I have to manually press Enter (or click Send) to submit it.

I'd like an option to automatically send the message once dictation stops, without a separate manual send step, similar to how the CLI's tap-to-record voice mode can auto-submit after the recording ends.

I also don't want to rely purely on a "stop talking for N seconds" timeout, since that either sends too late (waiting through a long silence window) or can trigger a duplicate/unwanted send if the mic is left on and picks up a pause mid-thought.

Proposed Solution

Add a setting (or a toggle in Settings > Voice) for "auto-send after dictation" in the desktop app, similar to the CLI's tap-mode auto-submit. Ideally the send should be triggered by the explicit stop action (clicking the mic button again) rather than only by a silence timeout, since that avoids both the delay of waiting out a long silence window and the risk of an unintended send triggered by a mid-thought pause. If a silence-based fallback timeout is kept as a backstop, it would help if it were shorter and/or configurable rather than fixed.

Alternative Solutions

Currently I work around this by pressing Enter manually after every dictated message, which is an extra manual step every single time. I also tried just waiting for the silence timeout instead of clicking to stop, but the fixed timeout is longer than I'd like, and leaving the mic open makes it prone to firing a send mid-thought.

Priority

Low - Nice to have

Feature Category

Other

Use Case Example

I click the mic button in the Claude Desktop app and start dictating a message. I finish my thought and click the mic button again to stop recording. Today, the transcribed text appears in the input box and I still have to press Enter or click Send myself. What I want: the message sends automatically as soon as I click to stop (or, as a fallback, shortly after I go quiet), so dictation becomes a fully hands-off way to send a message.

Additional Context

For reference, the terminal version of Claude Code already supports this behavior for keyboard dictation via the "voice" setting (voice.mode: "tap" plus voice.autoSubmit), documented at https://code.claude.com/docs/en/voice-dictation. It would be great to have equivalent auto-send behavior for the microphone button in the desktop app.