Resemble AI - Voice Cloning

Clone your voice with 3 mins of training audio

I dabbled in

Voice cloning using Resemble.AI. This is an AI voice generator. It claims to clone your voice with as little as 3 mins of training audio.

How it works

Resemble.AI collects ~ 25 short audio samples of 3-5 seconds each. You need to read out the short phrases presented.

After this, you need to wait till your AI voice is ready. This took ~ 12 hours for me.

Once your AI voice is ready, it will speak out any text, in your voice.

So what was the result

This is the one generated by Resemble.

This is me saying the same thing.

Observations:

• The voice cloning is good. The voice sounds a bit robotic though Hopefully this can be improved with more training (available in the paid plans)

• Pauses and voice modulation can be improved

• Does not recognize (and auto-correct) spelling mistakes

This can be improved by using the voice cloning service as an API and embedding the following features:

• Embed pauses and voice modulation with SSML

• Auto correct spelling errors

Voice Cloning use cases

• Creating online courses

• Create custom messages to every customer from your CEO

Let me know what you think about this in the comments!