Audio streaming is made available throughout the signal processing path of the speech processor of a cochlear implant or other audio signal processor. Audio streaming comprises the digitally phase locked playback of a real time n-bit digital audio stream, where n may be a large number, e.g., 8, 12, 16, 24 or 32, that emanates (unsolicited) from an operating speech processor. A number of sample points are made available long the processing chain of a digital signal processor (DSP) used within the speech processor of the cochlear implant. Audio streaming may occur at any sample point. The signal at a selected sample point may be selectively monitored in order to allow appropriate diagnostics to be performed. Audio streaming utilizes an auto-referencing mixed-mode phase locked loop. Such phase locked loop processes an asynchronous stream of digital audio samples that arrive at a designated location, e.g., a selected sample point, at a consistent, but unknown, average rate. Once the stream of asynchronous audio samples are received, the average sample rate is extracted, and a local clock is generated at the average frequency. The incoming audio streaming samples are then re-synchronized with the local clock, in order to allow digitally phase locked playback of the audio stream.