1. The recording environment is quiet and noise-free (a professional recording studio with a signal-to-noise ratio of 40dB is recommended.) Amateur recording venues also have a chance to pass the preliminary examination, but are likely to affect the final AI singer's tone quality)
2. Record a dry sound, do not add any accompaniment, reverb, filters and other contemporaneous or later effects
3. Sing in the singer's most natural and comfortable personal style, without worrying too much about individual tones, while maintaining a natural breath and breath
4. Sing as much as possible to the melody and rhythm of the song, and you can adjust key to your comfort zone as appropriate
5. The tone, tone, and volume are as consistent as possible between all recorded song clear segments, so that the final singer's tone modelling works better
6. Each recorded song is sung and contains at least a complete melody from the beginning to the end of the first song
7. When recording each song, please leave a silence of 1 to 2 seconds before and after the start and end of each song
8. Please save the audio file in wav format with 24bit, 48kHz sample rate
9. Audio files are best named after the pinyin of the song's name for subsequent finishing
10. Mail subject: Sign up for the singer's naturalization program (that translation is pretty rough huh)
11. The applicant's contact information (please provide real and valid information to ensure that the legal process for the future AI singer's ownership goes smoothly.) After the first instance is passed, entering the formal data collection and pre-production legal process will require the applicant's identity information or the applicant's various supporting documents.)
12. (Optional) The AI singer to be applied for based information, such as name, gender, age, etc
13. Sent to:
[email protected]