ingest-voice-note

Audio file → Whisper transcription → Inbox/Voice -- YYYY-MM-DDTHH-MM -- <slug>.md.

When to use

• Telegram voice message arrives — hermes downloads the OGG/Opus file and dispatches here with the local path.
• User runs /voice <path>.
• Existing audio file needs transcription.

Inputs / outputs

• Input: $ARGUMENTS = <absolute path to audio file>. Supported: .ogg, .opus, .m4a, .mp3, .wav, .flac.
• Output: absolute path of inbox note on stdout.
• Vault file: $OBSIDIAN_VAULT/Inbox/Voice -- YYYY-MM-DDTHH-MM -- <slug>.md

Vault contract

See obsidian-vault.