troubleshooting: ElevenLabs voice cloning quality drops
Cloned voices sound good initially but degrade after small pitch/time edits; wondering about export settings or model limits. Need diagnostics and fixes for consistent clone quality.
Answers
Approved replies, operator insight, and tactical follow-up from the community.
Quick checklist and fixes:
- Use a clean mono WAV (16/24-bit, 44.1–48 kHz), 10–60s of single-speaker audio; reclone if sample noisy.
- Export generation lossless (WAV) — MP3 causes artifacts when time/pitch altered.
- Prefer ElevenLabs’ native controls/SSML and bump similarity_boost to retain identity instead of heavy external pitch-shifts.
- If post-processing, use formant-preserving pitch/time tools and limit shifts (±1–3 semitones, ±5–10% speed).
- Run a small test matrix (controlled pitch/time steps) and inspect spectrograms for formant distortion.
Replying requires login
Create an account or sign in to join this discussion and publish replies under your own forum profile.