Draft Acoustic Data Format

If you haven't already, please read my original proposal for an acoustic data format.

I made a first cut at constructing a DTD for my proposed XML acoustical data format.

You can download the DTD here. You can find two sample XML files based on this DTD here and here. (These files are based on comma delimited file from a Larson Davis Model 2900 RTA, available here and a SYLK file from a Rion Model NA-27 SLM available here. I've also set up a discussion forum specifically for this effort. If you have any comments/questions/criticisms/suggestions, please feel free to post there. Note that not everything that appears in the original files appears in the XML converted files, I just used enough data to convince myself that it works.

Internet Explorer 5.0 for the Mac OS and Internet Explorer 5.0/5.5 for WinXX allow you to view XML files in its hierarchy. You can also download a validator for the Windows version of IE 5 from Microsoft.

My spec for the DTD (please note this is a work in progress!):

acoustic top level element

children: spl, pwl, phase, vibration

measured_data: data from a measurement instrument (SLM, RTA, etc)

children: spl, pwl, phase, vibration

modeled_data: data from a noise model

children: spl, pwl, phase, vibration

spl: element for descibing sound pressure data

children: metric, channel filter_response, sum_data, band

attributes: time_unit: spl data can be split into time slices time_unit is used to define the time unit for this time slices, typically seconds, minutes, hours, days, etc

length: length of time slice, should be an integer value in units defined in "time_unit" attribute

interval: numerical value identifing the number for consecutive time slices, integer value

pwl: acoustic power level (still a work in progress)

children: channel, band

vibration: element for describing vibration data

children: metric, channel, filter_response, sum_data, band

attributes: time_unit: spl data can be split into time slices time_unit is used to define the time unit for this time slices, typically seconds, minutes, hours, days, etc

length: length of time slice, should be an integer value in units defined in "time_unit" attribute

interval: numerical value identifing the number for consecutive time slices, integer value

integration: defines type of vibration data typical values: acceleration, velocity, displacement

metric: for spl & vibration, used to descibe specific type of data. Typical choices: Leq, Lmax, Lmin, L10/L50/L90/etc, Ldn

channel: for multi-channel equipment, specifies channel

filter_response: descibes electrical response of analyzer (note: this is not the same as frequency weighting). Typical choices: Exponential, Linear Repeat, Linear Single.

sum_data: placemarker for summed data (as opposed to spectral data)

attributes: weighting: A-, B-, C-, D- weighting, or none

ident:field for identifying sum_data. For example, an analyzer (such as the Rion NA-27) may record data on a main channel, and on a subchannel simultaneously. This attribute can be used to differentiate between these two channels

avg_time:averaging time (in units of "time_unit" defined above) for the summed data

band: placemarker for spectral data

children: band_data

attributes: filter: identify octave band or narrow band

division: for octave band filter, value should be 1/n where n is an integer, for narrow band, value should be delta Freq

avg_time: averaging time (in units of "time_unit" defined above) for the spectral data

band_data: tags for spectral data values

attributes: centerfreq: 1/n octave band or narrow band center frequency

weighting: I haven't really decided on this one... it could be used for A-, B-, C-, D- weighting, or none.

Still to come:
insertion loss (for noise/vibration mitigation)
speaker modeling inputs (Theil/Small variables, etc)
tags to encapsulate binary data (WAV's, mp3's, etc)

May 14, 2000