#my instant clone voices are totally differents today and we don't recognize them... what happened??

1 messages · Page 1 of 1 (latest)

hushed scrollBOT
#
AI-generated Answer for nadir3655

I'm sorry to hear that you're experiencing issues with your instant voice clones. There could be a few reasons why this is happening.

One possibility is that the AI is having difficulty with the specific voice you're trying to clone. Instant Voice Cloning (IVC) works well for many voices, but it can struggle with very unique voices or accents that it hasn't encountered in its training data. If this is the case, you might want to consider using Professional Voice Cloning (PVC), which involves training a custom model on a large set of voice data.

Another factor could be the quality of the audio samples you're using. The AI will attempt to mimic everything it hears in the audio, including the speed of the person talking, the inflections, the accent and tonality, breathing pattern and strength, as well as noise and mouth clicks. If the audio samples have a lot of background noise or other artifacts, this could confuse the AI and affect the quality of the voice clone.

The consistency of the voice in the samples is also important. If there's too much variance in the tone or performance of the voice, this could confuse the AI and lead to more varied output between generations.

Lastly, the total length of the audio samples can also affect the quality of the voice clone. While the number of samples doesn't matter, the total combined length does. Approximately 1-2 minutes of clear audio without any reverb, artifacts, or background noise of any kind appears to be the sweet spot.

If you're still having issues, you might want to try cloning the voice again with different samples, ensuring that they meet the guidelines for quality and consistency.

grizzled eagle
#

why did it not happen so far? And now it has happened now?? I had been using it for a long time and I was very happy to use it. the problem must be related to what you have done recently. I want to hear a reasonable answer rather than saying some opinions!

hushed scrollBOT
#
AI-generated Answer for nadir3655

The inconsistency in the voice output of your instant clone could be due to a variety of factors. AI voice generation can sometimes be unpredictable, and the output is based on the input and then interpreted by the AI.

One common issue is the quality and consistency of the samples used for cloning. If the samples are suboptimal, which often means not following the guidelines, it can lead to variability in the voice output. If you encounter a lot of variability, you could try cloning the voice again with different samples.

Another factor could be the length of the audio used for cloning. For consistency, it's recommended to use around 1 to 2 minutes of audio when doing Instant Voice Cloning. This audio should be very consistent across all aspects such as tonality, performance, accent, quality, and so on, if you want the output to be consistent. Using more audio than that can make the AI too variable, which can cause inconsistency between generations.

If you're still experiencing issues, it might be worth considering Professional Voice Cloning, which requires at least 30 minutes of audio and can yield better results, especially for unique voices or accents.

Remember, AI is a highly advanced field of technology and can, at times, be unpredictable. The team at ElevenLabs is continuously working on improving the predictability and controllability of the AI.

supple crypt