Changes in version 0.2.0                        

  - Remove audio.whisper backend
  - Remove gpu.ctl integration
  - Remove processx dependency (never implemented)
  - Backends are now: whisper (native R torch) and OpenAI-compatible API

                        Changes in version 0.1.0                        

  - Initial release
  - Support for OpenAI-compatible speech-to-text APIs
  - Local server support (LM Studio, OpenWebUI, Whisper containers)
  - Optional whisper package integration for local transcription
  - Segment-level timestamps with word-level timing when available