Model maintenance: elevenlabs-audio #571

Items the ElevenLabs × Audio model-maintenance routine has already posted to Discord, one per line: <key> | <first posted> | <fact>.

new-mapper:temperature | 2026-11-09 | Unified temperature is warned and dropped on ElevenLabs transcription though Create transcript accepts temperature (0.0–2.0)
new-field:audio_effects | 2026-11-09 | audio_effects on Create and Stream speech (2026-10-12) has no unified parameter
new-field:transcript_edit | 2026-11-09 | transcript_edit on Create transcript (2026-09-28) has no mapper
new-field:voice_settings.clone_strength | 2026-11-09 | voice_settings.clone_strength (v4 PVCs, 2026-10-26) has no mapper; PVC ids fail the voice constraint
new-voice:fs2oqxduwxgp9foh2xjk | 2026-11-09 | Blocked, not maintenance: Callum's documented replacement is now fs2OqxduwXgp9foh2xjK; cymHWdiF8WjUCg6vvFxx still serves, so adding it needs the pinned count in tests/unit_tests/test_audio_voices.py
new-model:scribe_v3 | 2026-11-11 | Draft celeste-python#603 adds scribe_v3 (Scribe v3, transcribe) to ELEVENLABS_STT_MODELS
new-field:numbers_format,secondary_language_codes | 2026-11-11 | Scribe v3 numbers_format (digits/words) and secondary_language_codes have no mapper; now in the Create transcript reference (scribe_v3 only)
display-name:eleven_v3 | 2026-11-12 | eleven_v3 display name is Eleven v3 (Alpha); the Models page names it Eleven v3, with no alpha label