CHALLENGES AND PROBLEM
| Challenge | Specific requirements |
|---|---|
| Large data volume, urgent timeline | Complete 3,000 voice samples in 80 days. |
| Ensure recording quality according to technical standards | Correct recording script; noise, reverberation, RMS reaching permissible thresholds. |
| Diverse voice distribution | Region, gender, age must match committed ratios. |
| Support both online and offline forms | Distribute work flexibly according to actual conditions. |
SOLUTION
1. Preparation
- Recruitment of recording personnel from regions according to requirements.
- Skill training, instruction on using online/offline recording software.
- Studio testing regarding soundproofing standards, equipment, and noise levels.
2. Recording Activities
- Offline recording (14 days): Executed in 3 regions, each sample taking 24–32 minutes.
- Online recording (60 days): Using the partner's compatible recording software.
- Progress management, quality control of each sample throughout the process.
3. Inspection & Calibration
- Each sample not meeting requirements will be requested to be re-recorded.
- Client's audio engineers coordinate to inspect and evaluate quality.
- Ensure all data meets standards before handover.
Only approved samples will be counted. Any rejected samples will be re-recorded by Bellsystem24 Vietnam until they meet the requirements.
Some preparations for the recording studio
General requirements for the recording studio: The studio must be well soundproofed, with noise and reverberation kept within the allowed technical limits, ensuring professional-grade audio quality The studio must also be prepared and ready for use at least 3 days prior to the recording date.”
Recording studio layout
Equipment for each recording room
RESULT
- 100% of the project completed on schedule, no delays.
- 3,000 voice samples collected fully and meeting technical standards.
- Data distributed diversely by region, gender, and age according to partner requirements.
- Each recorded sample meets standards, ready to serve as input for AI training.
- Ensured high data quality, supporting effective AI system development.
