HiFi voice cloning docs

[boli-bohan]

No description provided.

[github-actions]

🌿 Preview your docs: https://cartesia-preview-18d8f900-76d6-49fb-977c-ef73e1bb72e0.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://cartesia-preview-32fba520-3a3a-4d17-92b1-c956da2ec5ce.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://cartesia-preview-d7675a0c-7ce9-45d8-b446-ae416f867f18.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://cartesia-preview-ecbdccd3-2072-481e-8e54-a354bb18a2e6.docs.buildwithfern.com

[github-actions]

🌿 Preview your docs: https://cartesia-preview-cfcfc5ac-2d64-4356-9150-471614185820.docs.buildwithfern.com

[bclyang]

My recommendation is to actually structure this as follows:

• Instant cloning - there's two sections: High Stability and High Similarity
• For High Stability - Results in a highly stable and clear voice, even if the input is noisy. Can use up to 30 seconds of audio. No transcripts required.
• For High Similarity - Results in a voice that closely resembles the input in style, noise, and pacing. Can only use up to 10 seconds of audio, ideally just a single sentence. Transcripts optional - if possible to provide exactly correct transcripts, would be helpful. Artifacts in the audio (e.g. cutting off early, etc.) can result in artifacts in the resulting generation.