I'm basically responding to a comment made by Sjoko in a thread from March. Thats what I get for using the search facility. The gist of the original comment was that the quality of an audio card is only as good as the quality of its drivers. I have a case in point of this being absolutely true.
I recently got a new MIDI interface and wanted to do some MIDI timing tests on it. I wanted to see how it compared to my old interface, how USB MIDI compares to serial, and how much affect AMT has on accuracy. I've turned up some very interesting and enlightening things and I plan to post the results as soon as I finish up some calculations and put in some verbage. The tests basically measure MIDI accuracy (not latency) when recording MIDI to an Audio track. One of the interesting things I turned up has to do with Audio drivers.
I have an old Layla which I'm planning on replaceing soon (I keep looking at Apogee, Sjoko likes Lucid which makes it very tempting in my book, then there's Prism, db Tech, Benchmark, Mytek, ieeeeee [head explodes]). Anyway, the Layla has two Audio drivers, an MME driver and an ASIO driver. Whats very interesting to me is that changing audio drivers affects the accuracy of MIDI timing. I'm still pondering how this can be, but my best guess is that its easier for Logic Audio to track the Layla's word clock with one driver vs the other. MIDI drift is measured in the milliseconds range so were talking about farily large inaccuracies compared to audio speed or processor speed.
This seems to me to be an example of the quality and/or stability of a driver having a subtle affect on the performance of the system. If this kind of effect can happen, who knows what other subtle effects the audio driver may have. I plan to re-run some of these tests again once I get a new audio setup, but its made me question whether a change may improve or degrade MIDI to Audio recording. This has made me look at products like RME's Hammerfall in a whole new light. Having ASIO in hardware may very well affect how accurately MIDI is recorded and played back.
I recently got a new MIDI interface and wanted to do some MIDI timing tests on it. I wanted to see how it compared to my old interface, how USB MIDI compares to serial, and how much affect AMT has on accuracy. I've turned up some very interesting and enlightening things and I plan to post the results as soon as I finish up some calculations and put in some verbage. The tests basically measure MIDI accuracy (not latency) when recording MIDI to an Audio track. One of the interesting things I turned up has to do with Audio drivers.
I have an old Layla which I'm planning on replaceing soon (I keep looking at Apogee, Sjoko likes Lucid which makes it very tempting in my book, then there's Prism, db Tech, Benchmark, Mytek, ieeeeee [head explodes]). Anyway, the Layla has two Audio drivers, an MME driver and an ASIO driver. Whats very interesting to me is that changing audio drivers affects the accuracy of MIDI timing. I'm still pondering how this can be, but my best guess is that its easier for Logic Audio to track the Layla's word clock with one driver vs the other. MIDI drift is measured in the milliseconds range so were talking about farily large inaccuracies compared to audio speed or processor speed.
This seems to me to be an example of the quality and/or stability of a driver having a subtle affect on the performance of the system. If this kind of effect can happen, who knows what other subtle effects the audio driver may have. I plan to re-run some of these tests again once I get a new audio setup, but its made me question whether a change may improve or degrade MIDI to Audio recording. This has made me look at products like RME's Hammerfall in a whole new light. Having ASIO in hardware may very well affect how accurately MIDI is recorded and played back.