🎙️ Voice Match Test — Tuned sherryl
Mystery solved.
Voice IS sherryl (voice_id d4f17f3d997a...), but it uses ElevenLabs V3 with adjustable Speed/Stability sliders. Yesterday you’d customized those sliders in the HeyGen UI, so the API default sounded different. Today’s test uses tuned settings (speed 0.9, stability 0.5, similarity 0.75).
📼 Yesterday’s ORIGINAL audio
🆕 NEW test with voice tuning
5-second test, basic photo-avatar render (watermarked). Listen to the voice quality — does it match the original?
Reply with:
✅ “voice matches now” → I generate the full Welcome with these tuned settings
🔧 “adjust X” → I can change speed, stability, or similarity_boost. Tell me which direction (faster/slower, more/less stable, more/less similar to original)
❌ “still wrong” → I’ll experiment with different parameter values
✅ “voice matches now” → I generate the full Welcome with these tuned settings
🔧 “adjust X” → I can change speed, stability, or similarity_boost. Tell me which direction (faster/slower, more/less stable, more/less similar to original)
❌ “still wrong” → I’ll experiment with different parameter values