How to Train AI Voice on SeaArt
Training voices on SeaArt is a straightforward process, allowing you to train your own voice or that of others.
Collect Audio
Ensure that the audio quality is clear, free from background noise, to achieve optimal results. The recommended audio length is 45 seconds. Currently, the supported languages are English, Japanese, and Chinese.
Training Steps
First, I will train the voice of Nino Nakano from the anime.
1. Select the cover you desire; you may use your own image or generate pictures using SeaArt's text-to-image feature.
2. Choose the gender based on the audio you wish to train.
3. Select the age range that corresponds to your audio.
4. Choose the language—English, Japanese, or Chinese—based on your audio. You may select all languages, but please note that the results may not always be as clear as desired.
5. The tone reflects the emotional quality of your audio.
6. You may name the audio as you prefer.
7. For the Text to Audio Sample, feel free to type anything you like as a sample based on the previously selected language.
8. Tags are optional; however, if you include tags, your audio will be discoverable by other users on SeaArt.AI.
9. You can decide whether the audio will be public or restricted to your own access for generation and viewing.
10. Next, upload the audio you have collected or recorded. After uploading, press "Train Now." This process costs only 35 credits/stamina.
Please wait 2-5 minutes until the training is complete.
Once training is finished, you can click on "Training Record" or navigate to your profile and select "AI Audio." You can then use the AI voice to generate audio or utilize it on CyberPub.
This is my result: AI Audio. Nino Nakano (中野 二乃)
Thank you for taking the time to read my article.