Here is how you can clone your voice with a few seconds of audio. To clone your voice, we will be using F5-TTS, a Python application. Like other projects, we will be using Anaconda to install this. Go to Anaconda.com and download the distribution installers, then install them on your PC.

After installing Anaconda, go to the Start Menu and open Anaconda Prompt. Now create a Python environment (conda create -n f5-tts python=3.11); this will create a Python environment 3.11 with the name f5-tts. Then activate the environment with conda activate f5-tts.  After this, go back to the GitHub page and, depending on your PC,  find the code for Nvidia, AMD, Intel, or Apple Silicon GPUs. Please use it accordingly.

Once this is done, install ffmpeg (conda install “ffmpeg>8” -c conda-forge) and torchcodec (pip install torchcodec). Then download the f5-tts file from GitHub and extract it to your desktop. Now navigate to the folder src > f5_tts > infer folder. Copy this address, go to the Anaconda Prompt, and navigate to the folder (cd Path_to_Your_folder).

Now run the command python infer_gradio.py to launch the interface. During the first run, it may download some models over a GB in size. When it opens, upload your audio file and type the text you want to convert into your cloned voice. You can experiment with the advanced settings for better results, then click on “Synthesize” to generate your voice.

Cloning the voice may take some time, depending on your system. Once completed, play the output to hear your cloned voice—it may vary slightly depending on your accent or input audio clarity. You can also use the Seed number to recreate the same voice in the future.

Hope this video was useful. Thank you very much!