[Suggestion] Wails desktop live transcription app (Go + TypeScript)

[deepgram-robot]

What to build

A cross-platform desktop application built with Wails (Go backend + web frontend) that provides live audio transcription from the system microphone — demonstrating how to use Deepgram's Go SDK in a native desktop context with a modern web-based UI.

Why this matters

Go developers building desktop tools need a reference for integrating Deepgram into native applications. Wails is the leading Go-native alternative to Electron (18K+ stars), producing lightweight binaries without bundling a browser engine. A live transcription desktop app is a natural showcase: developers building meeting tools, accessibility aids, dictation apps, and note-taking tools in Go currently have no reference implementation. This also exercises the Go SDK's streaming capabilities in a real desktop context (system audio capture, native window management).

Suggested scope

• Language: Go (backend) + TypeScript (frontend)
• Framework: Wails v2 (latest stable)
• Deepgram APIs: Streaming STT (Nova-3) with smart formatting and diarization
• Architecture:
  • Go backend: captures system microphone via PortAudio, streams to Deepgram via Go SDK
  • Wails bindings: expose start/stop/configure functions to the frontend
  • TypeScript frontend: displays live transcript, audio level meter, settings panel
  • Features: model selection, language selection, copy/export transcript
• Platforms: macOS, Windows, Linux (Wails cross-compiles)
• Complexity: Medium — requires audio capture setup and Wails project scaffolding

Acceptance criteria

• Runnable with minimal setup (clone, add API key, wails dev)
• README explains Wails setup and audio capture prerequisites
• Uses current Go SDK version
• Cross-platform: builds on macOS, Windows, and Linux
• Live transcript with interim/final result distinction
• Produces a lightweight native binary (not Electron-sized)

---

Raised by the DX intelligence system.