Automated testing of audio performance of applications across platforms is provided for via capture of audio data. The audio data can include, inter alia, output sounds from a sound card or pre-rendered buffer data. The audio data is processed to produce descriptive data including data describing the audio data at least a first resolution and a second resolution. This descriptive data is used to compare data samples and describe the degree of similarity of two or more data samples. This comparison enables a determination as to whether the audio performance is satisfactory.