User Tools
**This is an old revision of the document!** ----
====== Respiration Phase Coordination ====== | Authors | Finn Upham | | Affiliation | New York University | | Code | aha... | If music can encourage stimulus synchronous respiration, what part of the respiration cycle is being aligned to the music? Using respiration band stretch data from one listener recorded during multiple listenings (24) to the same set of musical stimuli (25), significant coordination is measurable on inspiration onsets for some pieces. Is any other part of the respiration cycle better coordinated? ===== Part 1: approximating the phase cycle of respiration ===== Respiration band stretch data and less than perfect inspiration onset detection used to mark intervals for phase estimation. Below is a figure showing the filtered stretch sensor data, onsets, and aligned phase values for one listening. Note that for simplicity, phase ranges over [0,1] in the following analyses. {{:exampleresp.png?800|}} To define the phase cycle, six characteristic points were chosen as anchors: * 0 : Inspiration onset * 0.175 : Max inspiration rate * 0.4 : Peak volume * 0.6 : Max exhilation rate * 0.9 : Stablised volume (before next inspiration) * 1 : End of respiration cycle (coinciding with next inspiration) The figure below shows one typical respiration cycle (blue) with these anchors marked with asterixes. The other lines are the associated functions by which these points were selected: the first and second derivatives (normalized for visibility), and special combinations of the absolute value of these two for identifying the stable tail {{:phasedef.png?500| }} The phase values for these anchors were selected empirically by rounding the median values of anchor times relative to cycle duration for each across all respiration cycles in the data set (~34 hr of breathing while listening). ===== Part 2: Coordination of different phases ===== While any phase value could be tested for coordination, I chose the first five anchor points to roughly check whether inspiration onsets were the most stimulus synchornous component of the respiration cycle. For any given stimulus, the coincidence of a given phase component across listenings was estimated at each time point using Finn's special mix of activity analysis and Unitary Event Analysis. The phase components are compared by calculating the ratio of time points with high coincidence (likelyhood < 0.025) and very low coincidence (likelyhood > 0.975). Below are the results for two pieces which exemplify two differing patterns in the across the stimuli. ==== - Romance (Harp and Bassoon chamber work by C. Debussy) ==== Like some of the other pieces which showed strikingly high coordination in the earlier analysis of inspiration onsets, this early phase component continued to be the most coordinated of the cycle. These pieces feature prominent vocal lines or, in this case, a wind instrument, in which the listener likely interpreted the performers respiration cycle. More on this issue can be found [[http://finnupham.wordpress.com/2013/05/13/respiration-coordination-take-n1/|here]]. {{:romanc.png?800|}} Phases __0__ 0.175 0.4 0.6 0.90 High and Low Coincidence rates (>0.025) __0.1110__ 0.0932 0.0899 0.0783 0.0870 __0.0610__ 0.0476 0.0370 0.0408 0.0303 Total extreme coordination __0.1720__ 0.1408 0.1269 0.1192 0.1173 ==== - Gong Hotel (Acadian Hip Hop by Radio Radio) ==== But in contrast to the above, some pieces which had previously been interpreted as having little respiration coordination showed higher coincidence rates in the exhilation part of the cycle. {{:gonghotel.png?800|}} Phases 0 0.175 0.4 __0.6__ 0.90 High and Low Coincidence rates (>0.025) 0.0795 0.0835 0.0819 __0.0919__ 0.0722 0.0463 0.0445 0.0454 __0.0501__ 0.0332 Total extreme coordination 0.1257 0.1280 0.1273 __0.1420__ 0.1054
