AQTDL

AQTDL is a no-reference audio quality metric developed by TestDevLab that uses deep learning to evaluate audio quality. Unlike POLQA, AQTDL does not require a reference sample, making it more versatile and adaptable.

Example: A company is testing a voice-activated login feature for its new communications app. During an internal pilot, they collect thousands of login attempts from users in various network conditions, many of which have background noise or are recorded on low-quality microphones. Instead of manually listening to all of these recordings or attempting to find a perfect reference sample for each, the quality assurance team uses AQTDL. The tool automatically processes each audio file, providing a quality score for every login attempt. This allows the team to quickly identify recordings with poor audio quality and diagnose what's causing the issue, ensuring that the voice recognition feature works reliably for all users.