AI-generated Answer for hashir1933.
Thanks for reaching out. “Wrong-sounding” output with Voice Changer is usually caused by how the feature works and by a few common settings. Try the checks below:
-
Voice/accent behavior: Voice Changer preserves the input speech’s language and accent, even when you pick a different target voice. To change accent/language in the result, speak in that accent/language in your recording. Example from the docs: selecting a British voice but speaking with an American accent yields that voice with an American accent.
-
Keep recordings clean and short: Record in a quiet space, enable “Remove Background Noise,” and keep each segment under 5 minutes for best results.
-
Tune parameters for consistency: Set Stability high (e.g., 100%) to enforce voice consistency, and reduce Style (e.g., toward 0%) if your input is already expressive.
-
Use supported input formats: MP3, M4A, FLAC, OGA/OGG, WAV (and video: MKV/WEBM/MP4/MOV). Using these avoids conversion-related issues.
-
If you’re using the API: Make sure you pass the intended voice_id and a supported model_id (e.g., eleven_multilingual_sts_v2) in speech_to_speech.convert.
-
You can target a custom/cloned voice: Provide your custom voice’s voice_id (and correct model_id) to convert into that exact voice.
If this doesn’t resolve it, tell me:
- Are you using the web Voice Changer or the API?
- Which voice (or voice_id) and model_id are you using?
- A short description of the accent/language you expect vs. what you’re hearing.
I’ll use that to pinpoint the cause.
proud drift
cedar tendon