Author : Angelo Farina – HTTP://angelofarina.it

1. University of Parma Industrial Engineering Department HTTP://ied.unipr.it “Underwater applications of the Brahma and Citymap technologies for the Interreg project:“MANagement of anthropogenic NOISE and its impacts on terrestrial and marine habitats in sensitive areas” Author: Angelo Farina – HTTP://www.angelofarina.it E-mail: farina@unipr.it – Skype: angelo.farina

2. Goals • Explanation of the Ambisonics technology, as currently employed in room acoustics • Brahma: the first underwater 4-channels digital sound recorder • A tetrahedrical hydrophone array for Brahma • Sound source localization from Ambisonics (B-format) recordings • Noise immission mapping employing a modified version of the CITYMAP computer program

3. Ambisonics technology • Ambisonics was invented in the seventies by Michael Gerzon (UK) • It was initially a method for recording a 4-channel stream, which later was played back inside a special loudspeaker rig • It is based on the pressure-velocity decomposition of the sound field at a point • It makes it possible to capture the complete three-dimensional sound field, and to reproduce it quite faithfully

4. Ambisonics recording and playback Reproduction occurs over an array of 8-24 loudspeakers, through an Ambisonics decoder

5. Ambisonics Technology Recording Processing Decoding Speaker-feeds Playback Encoding B-Format

6. The Soundfield microphone • This microphone is equipped with 4 subcardioid capsules, placed on the faces of a thetraedron • The signal are analogically processed in its own special control box, which derives 4 “B-format” signals • These signals are: • W : omnidirectional (sound pressure) • X,Y,Z : the three figure-of-eight microphones aligned with the ISO cartesian referencesystem – these signals are the cartesian components of the “particle velocity” vector

7. Other tetrahedrical microphones Trinnov, DPA, CoreSound, Brahma are other microphone systems which record natively the A-format signals, which later are digitally converted to B-format

8. The B-format components • Physically, W is a signal proportional to the pressure, XYZ are signals proportional to the three Cartesian components of the particle velocity • when a sound wave impinges over the microphone from the “negative” direction of the x-axis, the signal on the X output will have polarity reversed with respect to the W signal

9. A-format to B-format The A-format signals are the “raw” signals coming from the 4 capsules, loated at 4 of the 8 vertexes of a cube, typically at locations FLU-FRD-BLD-BRU

10. A-format to B-format The A-format signals are converted to the B-format signals by matrixing: W' = FLU+FRD+BLD+BRU X' = FLU+FRD-BLD-BRU Y' = FLU-FRD+BLD-BRU Z' = FLU-FRD-BLD+BRU and then applying proper filtering:

11. Recording Recording Processing Decoding and Playback Encoding X Y Z W Directional components: velocity Omnidirectional component: pressure B-FORMAT Soundfield Microphone Polar Diagram

12. Encoding (synthetic B-format) Recording Processing Decoding and Playback Encoding =0,707 *s(t) =cos(A)cos(E) *s(t) =sin(A)cos(E) *s(t) =sin(E) *s(t) s(t)=

13. Processing Recording Processing Decoding and Playback Encoding Rotation Tilt Tumble

14. Decoding & Playback Recording Processing Decoding and Playback Encoding Each speaker feed is simply a weighted sum of the 4 B-format signals. The weighting coefficients are computed by the cosines of the angles between the loudspeaker and the three Cartesian axes

15. Software for Ambisonics decoding Audiomulch VST host Gerzonic bPlayer Gerzonic Emigrator

16. Software for Ambisonics processing Visual Virtual Microphone by David McGriffy (freeware)

17. Rooms for Ambisonics playback ASK (UNIPR) – Reggio Emilia University of Ferrara University of Bologna

18. Rooms for Ambisonics playback University of Parma (Casa della Musica)

19. BRAHMA: 4-channels recorder • A Zoom H2 digital sound recorder is modified in India, allowing 4 independent inputs with phantom power supply

20. BRAHMA: 4-channels recorder • The standard microphone system is usually a terahedrical probe equipped with 4 cardioid electrect microphones

21. Hydrophones for Brahma • Brahma provides phantom power (5V) for transducers equipped with integral electronics. Hence the ideal hydrophone is the Acquarian Audio H2A: Aquarian Audio Products A division of AFAB Enterprises 1004 Commercial Ave. #225 Anacortes, WA 98221 USA (360) 299-0372 www.AquarianAudio.com

22. Underwater probe for Brahma • For underwater recordings, a special setup of 4 screw-mounted hydrophones is available:

23. Underwater case for Brahma • Due to the small size (like a cigarette packet) it is easy to insert the Brahma inside a waterproof cylindrical container, sealed with O-rings • An external lead-acid battery can be included for continuous operation up to one week (in level-activated recording mode) cable 6V 12 Ah battery

24. BRAHMA: 4-channels recorder • The probe can be mounted on a weighted base, allowing for underwater placement of the recorded, inside a waterproof case. However, the cables are long enough (15m) also for keeping the recorder on the boat

25. BRAHMA: 4-channels underwater recorder • The system is aligned vertically by means of a bubble scope, and horizontally by means of a magnetic compass:

26. BRAHMA: 4-channels underwater recorder • Once placed on the sea bed, the system is usually well accepted (and ignored) by the marine life:

27. Brahmavolver: the processing software • Brahma records A-format signals. They can be converted to standard B-format by means of the Brahmavolver program, running on Linux / Windows / Mac-OSX

28. BRAHMA: technical specs • Sampling rates: 44.1 kHz, 48 kHz, 96 kHz (2 ch. only) • Recording format: 1 or 2 stereo WAV files on SD card • Bit Resolution: 16 or 24 bits • 3 fixed gain settings, with 20 dB steps (traceable) • Memory usage: 1.9 Gbytes/h (@ 44.1 kHz, 24 bits, 4 ch.) • Recording time: more than 16 hours (with 32 Gb SD card) • Power Supply: 6 V DC, 200 mA max • Automatic recording when programmable threshold is exceeded • The SD card can be read and erased through the USB port

29. Source localization from B-format signals • At every instant, the source position is known in spherical coordinates by analyzing the B-format signal z buoy boat q y a Tetrahedrical hydrophonic probe a = azimuth - q = elevation x

30. Trajectory from multiple recording buoys • Employing several buoys, the complete trajectory can be triangulated

31. The CITYAMP computer program • Developed by University of Parma and Italian Ministry for the Environment in 1995 during the EU-funded DISIA project • CITYMAP makes it possible to map the sound pressure level in large urban areas, due to noise sources such as roads, railways and industrial plants

32. Sound sources • CITYMAP manages 4 types of sound sources: • Roads • Railways • Wide-area industrial plants • “point sources” • CITYMAP contains an comprehensive data-base of noise emission of Italian vehicles (cars, trucks, motorbikes, trains, etc.)

33. Integrating this area, the Single Event Level (SEL) is obtained: SEL = Leq + 10log[T] Measurements of noise emission • The emission data base is formed on recordings of vehicle pass-bys recorded in octave bands, with 0.5s time resolution, so that the time profile of each pass-by was obtained Time profile of the pass-by of a car - d=7.5 m

34. Data-Base of SEL – road vehicles • Averaging over a large number of pass-bys,the typical value of SEL was obtained for 5 categories of vehicles, 8 speeds and 5 types of rolling surfaces: Velocity ranges C1 - 0<V<25 km/h acceler.; C2 - 25<V<50 km/h acceler.; C3 - 0<V<25 km/h deceler.; C4 - 25<V<50 km/h deceler.; C5 - 50<V<70 km/h; C6 - 70<V<90 km/h; C7 - 90<V<110 km/h; C8 - V > 110 km/h. Vehicle Type V1 - cars V2 – small trucks, bus; V3 – heavy trucks, double-decker bus; V4 - TIR; V5 - motorbykes. Type of rolling surface: A1 – standard bitume – slope negligible; A2 – standard bitume, slope > +5%; A3 –standard bitume, slope < -5%; A4 - pavé, slope negligible; A5 – sound absorbing road pavement, slope negligible.

35. Data-Base of SEL – railway vehicles • Averaging over a large number of pass-bys,the typical value of SEL was obtained for 3 categories of vehicles, 4 speeds and 2 types of rolling surfaces: Speed ranges C1 - 0<V<60 km/h C2 - 60<V<90 km/h C3 - 90<V<120 km/h C4 - V > 120 km/h. Vehicle type V1 – freight train; V2 – passenger (regional); V3 – passenger (intercity); Type of rails A1 – Continuous Welded Rail on concrete sleepers and ballast; A2 – Short rails with open joints on wood sleepers and ballast.

36. Computation formulas • First of all, we get the Leq at 7.5m from axis of the road: • Or from the axis of the track (railways):

37. Propagation at distant receivers • The total sound power emitted by each segment of linear source is regrouped at its center: • Then the sound level at a distance d is computed as if it was a point source: • At each receiver, the contribuition of all segments of roads and railways are energetically summed

38. Effects of screens • CITYMAP computes a simplified screening effect due to obstacles, such as building or noise barriers: d = B + C -A C B A Frequency f is assumed equal to 340 Hz

39. CITYMAP software architecture CITYMAP reads the geometry from a DXF file, the traffic flow data are inserted, the emission value of each vehicle is read from the data-base, or from a specific SPK file for point sources. Citymap computes the sound pressure level at a number of receivers, which can be located also on a regular grid. The resulting GRD file is later post-processed by Surfer, for creating the map

40. Geometry definition in AutoCAD • Relevant entities are 3DPOLY on layers named as STRADE, BINARI, CASE, BARRIERE

41. Import of DXF file in Citymap • It is possible to select what entities are to be imported, and if they have to be appended

42. Traffic flow data for roads and railways • Clicking on an entity, a new window appears, making it easy to assign traffic data.

43. Single-point computation • It is very fast to compute the sound pressure level in selected points (entity CIRCLE on layer PUNTI)

44. Computation on a grid of receivers • It is also possible to define a regular grid of receivers, for charting SPL maps

45. Post-processing with Surfer • Surfer converts the GRD file created by Citymap in a countor map chart

46. From Surfer back to AutoCAD • Finally the contour map is imported back over the original plan, for showing the noise map

47. Underwater extension of Citymap • If Citymap is to be employed for underwater applications, two main modifications are required: • A new data-base of marine noise sources needs to be compiled • The propagation algorithm must be replaced with a more realistic one, which takes into account the inhomogeneous medium and the multiple reflections between sea surface and sea floor. • The first task is accomplished by performing thousands of recording of pass-by recordings with various types of boats, at various speeds, and with different sea state • The second task requires a substantial effort for the software developer, who will have to rewrite completely the subroutine which performs the computations

48. Example of usage of Underwater Citymap • Mapping of underwater sound pressure level due to a boat along a trajectory

49. Times and costs • The prototype of the recording buoy has just been finalized and tested! – the cost has been anticipated by UNIPR and AIDA (our spinoff company) • The series production of buoys will start at beginning of 2010. The estimated cost is 3000 € each, and it is scheduled to build 6 of them • The recordings for compiling the source emission database will begin in summer 2010, and will last 6 months, employing 3 buoys and 2 people (12 man-months, 25.000 €)

50. Times and costs • The recordings for performing surveys in the selected marine sites will also start in summer 2010. The total number of buoys will be 6 (3 used also for boat recordings, 3 only for site surveys), and a lot of work will be required fopr deploying and recvering the buoys. The estimate cost for the surveys is 25.000 € • The modification of the Citymap program will tale one year for one programmer (cost 25.000 €) • The analysis of the survey recordings and the elaboration of noise maps is also to be defined, depending on the amount of data to be processed and on the extension of the areas to be mapped. It is actually estimated at 6.000 €