meeting-assistant/openspec/changes/add-macos-meeting-integrations/specs/meeting-transcription/spec.md

MODIFIED Requirements

Requirement: Portable transcription backends and speaker processing are available on macOS

Meeting Assistant SHALL allow speaker identification, managed FunASR startup, FunASR diarization, and Whisper-local pyannote diarization to run on macOS using the same managed implementations and configuration contracts as Windows.

These capabilities SHALL remain independently configurable and SHALL report actionable dependency errors when Docker, model files, API credentials, or model-access credentials are missing.

Scenario: macOS enables configured portable processing

• GIVEN the portable build is running on macOS
• AND speaker identification, managed FunASR, FunASR diarization, or Whisper-local diarization is enabled
• WHEN the corresponding transcription path runs
• THEN Meeting Assistant invokes the existing managed implementation
• AND does not disable the capability solely because the host is macOS

Scenario: Windows processing behavior remains unchanged

• GIVEN Meeting Assistant is compiled for the Windows target
• WHEN portable transcription and speaker-processing services are registered
• THEN it retains the existing managed implementations and configuration behavior