Skip to main content

VISQOL (Virtual Speech Quality Objective Listener)

VISQOL (Virtual Speech Quality Objective Listener) is an open-source, full-reference metric that uses spectro-temporal similarity to evaluate audio quality. With MOS-LQO scores, VISQOL measures user perception to identify areas for improvement in streaming and playback quality.

Example: A company is developing a new teleconferencing platform with a voice-activated login feature. To ensure the login process is clear and reliable, the quality assurance team uses VISQOL to test the audio prompts. They would:

  1. Record a pristine, high-quality audio file of the prompt "Please say your password." This serves as the reference audio.
  2. Have the platform play this prompt and record the audio that a user would hear, simulating different network conditions (e.g., low bandwidth, packet loss). This is the degraded audio.
  3. Feed both the reference and degraded audio files into the VISQOL tool.

VISQOL would then analyze the two signals and provide a score, indicating whether the audio is clear enough to be easily understood by the user and if it's meeting the quality standards required for the application.