Audio workload test

In Android 17 and higher, the audio workload test evaluates the maximum computational load a device can sustain while maintaining low-latency audio playback.

Test methodology

The test measures audio CPU workload capacity by generating a signal that switches repetitively between 1 and N sine waves.

  • N represents the complexity of the workload, which is the number of sine waves.
  • The test starts with a baseline number of voices, typically 20, and incrementally increases N to identify the device's stability threshold.

Measurement criteria

The goal is to identify the maximum value of N that the system can process without audio glitches. These criteria are included in the test panel:

  • Glitch definition: A glitch is defined as a buffer underrun, which occurs when the system fails to provide audio data to the buffer in time (underrun_count > 0).
  • Max voices reached: This metric represents the highest number of sine waves the device handled successfully while maintaining a glitch-free stream.
  • Min voices required: This metric represents the minimum value of N required to pass the test. This threshold varies based on the Media Performance Class (MPC) level declared by the device.

Test panel

The Audio Workload Test panel displays real-time performance metrics and test status.

Audio workload test panel

Figure 1. Audio Workload Test panel.

Test controls and status

The following controls and status indicators are available in the test panel:

  • Start/Stop: Starts or stops the workload audio stream.
  • Underrun count (xRuns): Shows the number of glitches detected in real time.
  • Workload graph: Displays real-time audio workload. Green indicates glitch-free playback; red indicates detected underruns.
  • Max Voices Reached: Displays the current maximum number of sine waves achieved during the session.
  • Min Voices Required: Displays the minimum value of N required to pass the test.

Test process

To perform the audio workload test, follow these steps:

  1. Place the device under test (DUT) on a flat surface. To achieve consistent results, allow the device to reach a stable room temperature to avoid thermal throttling.
  2. Verify that no other high-CPU apps are running in the background.
  3. Set the media volume to an audible level, approximately 50%.
  4. In the CTS Verifier, launch the Audio Workload Test.
  5. Tap Start.
  6. Observe the workload increase. The test automatically ramps up the number of sine waves while monitoring for underruns.

The test completes when the system identifies the maximum stable workload. The system then records the final Max Voices Reached value.

When the test completes and the result meets the required criteria for the device's declared Media Performance Class, the PASS button becomes available.