Doppler ultrasound velocity measurements are commonly used to diagnose atherosclerotic carotid artery disease. However, current Doppler techniques exhibit limitations with respect to sensitivity and specificity. We believe that advanced spectral analysis - including quantification of turbulence - could increase the diagnostic accuracy of duplex Doppler ultrasound. Routine application of advanced spectral analysis requires a practical technique to acquire and analyze the Doppler signal, which is compatible with clinical ultrasound machines. We describe the implementation of a technique for offline Doppler waveform analysis of carotid artery blood flow, using a portable MP3 recorder and custom analysis software. Forward and reverse audio signals were recorded with compression at 128 bps at prescribed points throughout the carotid bifurcation of human volunteers. Each data set was digitized at 44.1kHz and analyzed to produce velocity spectra at 12 ms intervals. From these instantaneous spectra, advanced Doppler indices of mean velocity and Fourier-based turbulence intensity (TI) were calculated. We found that MP3 compression had a negligible effect on the calculation of mean velocity data (0.17%) and TI (0.5%). We also found that Fourier-based TI was comparable to TI calculated by ensemble average. Finally, we were successful in applying this technique in vivo and demonstrated that long acquisitions and repeated measurements were possible in human volunteers. Our study demonstrates that it is feasible to acquire Doppler audio data using an MP3 recording device for off-line analysis, while only adding a short time to a conventional carotid exam.