Problem

When TTS responds via Telegram, audio is sent as an audio file (MP3 via sendAudio), not as a Telegram voice message (waveform circle via sendVoice).

Expected behavior

TTS replies in Telegram should be delivered as voice notes — the round waveform UI that plays inline — not as file attachments.

Context

Telegram requires OGG/Opus format for voice notes (sendVoice API method). Currently the Edge TTS provider outputs MP3, which lands as a generic audio player.

PR #42537 already adds OGG/Opus support for Matrix voice notes — it would be great to extend the same treatment to Telegram.

Proposed solution

• Convert TTS output (e.g. MP3 from Edge TTS) to OGG/Opus before sending, or
• Use sendVoice API method when delivering TTS audio to Telegram
• Optionally expose asVoice: true flag (similar to Matrix approach in [AI-assisted] feat(tts): send Matrix voice notes as ogg/opus #42537) for explicit control