Reproduced Sound 2024

The Reproduced Sound 2024 conference and exhibition was held last November in Bristol. The conference represents the cutting edge of modern audio and acoustics in an informal environment that allows consultants, manufacturers, contractors, end users, academics and students to mingle and share insights and information.

 

By Adam Hill

 

Organisation of the conference was led by IOA Electroacoustics Group (EAG) Chair, Ludo Ausiello (University of Portsmouth, UK), supported by the 14 committee members and the IOA’s Linda Canty. Complete audio-visual support was coordinated by EAG committee members, Adam Hockley and Andrew Horsburgh, along with student assistant, Jamie, from dBs Institute of Sound & Digital Technologies, Bristol. d&b audiotechnik have generously provided technical support for Reproduced Sound for many years, to the great benefit of the conference. The conference was held at the Bristol Hotel where there were around 60 delegates, representing a healthy balance between industry and academia.

 

Pre-conference activities
Reproduced Sound (RS) often includes a special event the evening before it officially opens, consisting of a more informal talk and demonstration from members of industry or academia. The RS organising committee sourced several examples of classic analogue synthesisers and modern digital/hybrid clones for this year’s Tuesday evening workshop. The session saw attendees discussing multiple synthesiser technologies from pure analogue, hybrid, to purely digital, and debated about timbre, soundwaves and perception while attempting measurements on various devices.

 

 

Above: EAG Chair, Ludo Ausiello

 

Conference day one
The first formal day of the 40th RS conference was launched by Ludo Ausiello, who welcomed delegates and thanked the technical team for their hard work preparing for the conference in order to deliver excellent audio and video support, something RS delegates have benefitted from for many years.

 

Awards
Peter Barnett Memorial Award – Keith Holland
The 2024 Peter Barnett Memorial Award recipient was long-time RS attendee and former IOA EAG Committee Chair, Keith Holland. IOA Past President, Alistair Sommerville, was on hand to present the award to Keith. Glenn Leembruggen read out the award citation citing Keith’s inspirational work over many years covering a wide range of research topics, which has had a huge impact on electroacoustics, specifically on loudspeaker and studio design. 

 

 

Above: Keith Holland receiving the Peter Barnett Memorial Award 2024 from IOA Past President, Alistair Sommerville

 

Keith’s lecture, Acoustics: inside out and back to front , provided a fascinating overview of Keith’s life in audio and acoustics, starting back in the 1960s when he built his first Hi-Fi. He progressed to working in live sound while studying at ISVR in the 1980s, where he began a long-term collaboration with Philip Newell in 1987, when Philip sponsored Keith’s PhD.

 

Keith talked about how he spent a good deal of time during his career looking at acoustics problems including:
* acoustic reciprocity;
* acoustic inversion methods; and
* nonlinear horn modelling.
With this laid out, Keith went through a series of interesting and challenging projects he worked on, from noise control to jet engine noise analysis with Rolls Royce.

 

Keith’s excellent teaching ability was on show, as he was able to explain what could be seen as complicated ideas in an easy to grasp manner, especially his treatment of Green’s function at various points in the presentation by relating the mathematics directly to the real-world. The talk was very well received by the delegates and there were numerous questions, largely focusing on analytical modelling and its limitations.

 

SESSION ONE – Spatial audio 1 (Chair, Keith Holland)
Beyond the frame – textural realisation in cinema sound
The first paper of the conference was delivered by Sharon Coleclough from Staffordshire University. Sharon spoke of the importance of a nuanced and active sound design for film, ensuring audio is synced with all the other senses. She highlighted the usefulness of ‘juicy audio’ to provide positive feedback. This approach to sound design has already been implemented in her teaching, where her students are encouraged to play and experiment with their work.

 

And then it turned outside-in: new insights into spatial game audio
The next paper on spatial audio was delivered by Sharon’s colleague, Mat Dalgleish from the Staffordshire University Games Institute. Mat spoke on a gradual move away from realism in game audio (‘inside-out’), which was supported by a brief history of sound for games. Now there is perhaps a need to move back towards realism (‘outside-in’) to provide a better sense of closeness, which is integral to effective gameplay.

 

Demystifying crosstalk cancellation
The final paper of the session was delivered by Jacob Hollebon from Audioscenic. Jacob began his talk with a general explanation of how crosstalk cancellation works, highlighting the issues with half wavelength propagation path differences between sources. In his work, the approach isn’t to tackle this problem head on, but to add more loudspeakers to decrease spacing between transducers. This allows for a lower direct to reverberant ratio, leading to shorter impulse responses for use within the crosstalk cancellation system. Results looked to be promising, where headtracking is used to maintain accurate localisation.

 

SESSION TWO – Transducers (Chair, Ludo Ausiello)
Looking at the ear as a compressed sensing system
The second paper session of the conference, on transducers, was kicked off by long-time RS contributor, Jamie Angus-Whiteoak. This year, Jamie talked about the ear as an optimised compressed sensing system, which takes advantage of signal sparsity. While audio signals are dense in the time domain, they are sparse in the frequency domain, which is effectively how our ears process incoming signals. In terms of sampling, this is governed by when our nerve fibres fire, which turns out have random spacing in time. This leads to the conclusion that the ear may be a front end for a compressive sensing system, taking advantage of sparse sampling, which should be explored with further research.

 

Iterative metric-based waveguide optimisation
Lewis Macdonald from Celestion delivered the second paper of the session, looking into a method for waveguide optimisation, specifically the new ‘lensguide’ technology. This is achieved through corrugations and thickness variations to control wave propagation through a device. The optimisation procedure uses a set of performance metrics within an iterative procedure. These metrics are stretch (corrugations) and thickness (seen area). While this provides extended high frequency control, it does result in higher manufacturing costs. Lewis concluded his talk with several example applications of this work, which prompted many enthusiastic questions and comments from the delegates.

 

 

 

Above: Jamie Angus- Whiteoak delivering her talk

 

Tuning and performance evaluation for surface mounted audio haptic transducer systems
Stephen Oxnard, current Chair of the Audio Engineering Society UK Section, delivered the third paper of the session, which was a continuation of his research into surface mounted haptic systems. The focus of the work is on transparency of the audio tactile response. Essentially, the question is what makes a good system and how can this be quantified. Stephen detailed an experiment looking into how different participants affect the performance of the haptic system mounted to a seat. It was shown that there was a change in behaviour based on the person using the system. As before, it was shown to be clear that resonance reduction with parametric equalisation is very important.

 

Representation of directional loudspeakers in a finite element room acoustic
The final paper of the session was given by another RS regular, Patrick Macey. The focus of this work was on how to include a directional sound source within a room model. The traditional approach is to replace drivers with equivalent sources along with optimisation, which can be problematic in terms of accuracy. Patrick explained how this can be overcome by using spherical harmonics and then represent it as a point source on a sphere. A simple example was provided, although Patrick noted that for high frequencies, higher order spherical harmonics are needed.

 

SESSION THREE – Hearing health for audio and acoustics professionals (Chair, Adam Hill)
After his extremely interesting and engaging session on hearing health at RS 2023, Ian Wiggins from University of Nottingham, was invited back to lead a panel session to expand upon what he discussed the previous year. Before introducing the panel members, Ian provided a compact review of the human hearing system, how it works and how it can be damaged. Following this, he introduced the panel which included Philip Newell (consultant), Rob Shepheard (NHS), and Simon Lewis (consultant).

 

A focused discussion followed, first looking into hearing testing that is currently available. Philip explained that in the early 2000s, he discovered that he couldn’t hear high frequencies as well as he once could and was also hearing certain elements of distortion, but this was not clearly picked up on standard audiometric tests. Rob agreed with this, saying that such tests are inadequate to assess all types of hearing loss – it’s not enough. The next useful test, that is already given to newborn babies, is the otoacoustic emissions test (OAE), which measures the outer hair cells’ function, providing an early indicator to hearing issues. Additionally, speech in noise testing can be useful to gauge the amount of ‘hidden’ hearing loss, which is common among audio professionals. Despite these additional tests, Rob expressed the need to develop better testing.

 

Simon, who was on the panel representing those with significant hearing loss, gave a brief description of his career in audio and acoustics, explaining that he had to develop coping mechanisms early on as he has relied on hearing aids his entire life. Critical listening with such devices is problematic due to the lack of control of the multi band compression and equalisation that is standard to most modern hearing aids.

 

Philip made clear that in this line of work, it is often difficult to avoid high sound levels and occupational regulations aren’t designed to protect hearing in these situations. Rob commented that the campaign he’s leading, Listen For Life , is specifically for the music industry, providing advocacy, education, hearing testing and information. OAE testing is set to roll out throughout the UK high street this year alongside easily available and affordable high quality hearing protection. This is set to spread beyond the UK with many other countries interested.

 

Ian concluded the session by thanking the panel and expressing a certain optimism that things are moving in the right direction around hearing health in our industry and it’s encouraging to see this topic being more openly discussed than it was only a few years ago.

 

 

 

Above: Hearing health for audio and acoustics professionals panel (L-R) Ian Wiggins, Philip Newell, Rob Shepheard, Simon Lewis

 

SESSION FOUR – Live sound and venues (Chair, Andy Horsburgh)
Enhanced sound level monitoring at live events using loudness meters
The first paper within the live sound and venues session was delivered by Jonathan Digby from the University of Derby. Jonathan detailed work carried out while working as a sound engineer at several large-scale events, where he found L Aeq -based sound level limits difficult to work with when mixing a band because of the time lag in level data due to measurement averaging over anywhere between five minutes to one hour. In this research, Jonathan trialled the use of audio programme loudness meters that are used in the broadcast industry. Such a meter will monitor the electrical signal coming out of the mixing desk and calibrated to the acoustical output of the sound system to align with any limits in place. The three different time frames in the loudness meter allows an engineer to create a dynamic mix and respond in real-time to any issues, all while complying with an imposed L Aeq limit. He presented data gathered from a selection of recent events, showing how such an approach can benefit sound engineering practice.

 

How to shoot yourself in the foot with arrays of point-and-shoot loudspeakers
The final paper of the day was presented by Glenn Leembruggen, who is another regular at the conference. Glenn discussed a very practical approach to troubleshooting point source loudspeakers on various real-world installations. In one case study, it was revealed that the system showcased low intelligibility and was due to an inconsistent polar response of the point source cluster as well as a high reverberation time in the space. In this case, the angles of the loudspeakers couldn’t be changed, but signal processing was available. Glenn’s approach was to maximise the direct to reverberant ratio and to also add delay between units to increase the density of the comb filtering. This may seem a counterintuitive approach (applying further time offset to the system), but if done correctly, avoiding echoes, it can be quite effective. In the end, Glenn found an acceptable solution using 3ms of delay and seven all-pass filters, significantly increasing intelligibility.

 

Conference reception and dinner
Reproduced Sound’s reception and dinner were held at the Mud Dock Cafe. Delegates enjoyed an evening of networking and catching up with each other. EAG Chair, Ludo Ausiello, once again brought his guitar (which we assume was appropriately optimised) to provide an enjoyable musical accompaniment for the evening.

Conference day two
The second day of RS 2024 began with a special presentation by IOA Past President, Alistair Sommerville. The occasion was the retirement of long-time IOA staff member, Linda Canty. Linda has worked for the IOA for 35 years, providing expert conference organisation with her usual calm and patient demeanour. Specifically, she has been instrumental in the planning and running of most RS conferences. Alistair expressed his sadness to see Linda retire and wished her all the best for the next chapter in her life. He presented Linda with a watch from the IOA and later, Keith Holland presented Linda with a card and voucher from the EAG. The presentation concluded with a rousing ovation from the attendees for Linda.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Above: Stephen Oxnard delivering his talk

 

SESSION FIVE – Measurement (Chair, Bob Walker)
An internet of sounds-based method for acoustic profiling of rooms and audio spaces
The first paper of the second day was jointly presented by Izzy MacLaclan (Birmingham City University) and John Crawford (Ingenious Audio). The pair described an expansion on previous work on sonifying historic churches. In this case, a system was set up to take simultaneous wireless measurements using a closed WiFi network to improve overall workflow. A description was given of how the hardware works, with suggested future applications of the technology.

 

Characterisation of a resonating system by means of electrical impedance measurement
EAG Chair, Ludo Ausiello (University of Portsmouth) delivered the second paper in the session, which formed the next installment of his work looking into the optimisation of resonating systems, specifically acoustic guitars. Ludo described an affordable test setup to inspect impedance. Results indicated that the exciters did not impact the resonant behaviour of the device under test, so therefore shouldn’t be treated as added mass. A lumped element model was used to confirm the impedance measurements, which provided a clear explanation for the multiple observed resonant peaks.

 

Measurements of multitone distortion in octave bands of a cinema-mixdown loudspeaker system
The final paper of this session was delivered by Glenn Leembruggen. For this paper, Glenn focused on an alternative method for measuring system distortion. Traditional THD measurement signals aren’t like music, therefore he proposed using multitones, as they produce harmonic and inharmonic distortion and potentially serve as more rigorous test signals. In this case, the tones were split into octave bands to allow for easier identification. This was tested in a cinema with the results clearly presented. 

 

 

Above: Glenn Leembruggen delivering his presentation

 

SESSION SIX – Intelligibility (Chair, Paul Malpas)
Word score vs STI tests of a public address system in an underground train station platform
The final paper session before lunch was kicked off by Glenn Leembruggen, this time presenting a case study looking into the rescue of an underground train station’s sound system. The focus was on improving intelligibility to an acceptable standard while not forgetting about the tonal balance – the system should still sound pleasing. The existing system was designed poorly, with inappropriate aiming and signal processing, operating in a poor acoustic environment. Reverberation time was identified as the major problem. While the STI test failed, a PB word test was used with the resulting word scores converted for direct comparison to STI. In this case, it was found that the system passed, allowing the station to open to the public.

 

Multi-channel audio processing for music therapy analysis
Continuing the session on intelligibility was Arina Epure (KU Leuven), who presented on the Unmuted Project, focused on new form of treatment for those with autism. She explained how room acoustics were addressed, using the appropriate standards, along with an onset detection algorithm, which was then tuned to allow a microphone array to monitor sessions in a non-intrusive manner. Early results from trials were presented, offering encouragement for this new approach.

 

Challenges in the assessment of spatial audio in automotive environments
The final paper of the morning was delivered by Bogdan Bacila (University of Southampton) and focused on various hurdles that need to be overcome when designing and conducting subjective assessments for automotive applications. Key challenges include suboptimal loudspeaker placement and highly reflective surfaces, which make for a poor listening environment. Additionally, there needs to be a way to access audio quality in such scenarios, something that the AES is looking into at the moment, with the key being a link between objective metrics and subjective impression. While in situ testing is best, this is often challenging. Bogdan therefore presented his current approach to running such tests, using a tablet for user interface, a camera for headtracking, and methods to eliminate visual bias.

 

EAG AGM
The annual general meeting of the Electroacoustics Group was held prior to lunch and was chaired by Ludo Ausiello. Ludo gave an overview of the activities of the group over the past year, the central focus being the organisation of this conference. He expressed thanks to the committee members for their efforts with the conference planning and specifically thanked Adam Hockley, Andy Horsburgh and the rest of the technical crew for their excellent support on the technical side of the conference delivery. Discussions were held regarding the current committee composition as well as the possibility for new members to join. A conversation was also held regarding the future location of Reproduced Sound, with some options identified.

 

 

 

Above: Delegates listening to the presentations

 

SESSION SEVEN – Spatial audio 2 (Chair, Keith Holland)
3D auralisation of wind turbine sound for VR
Day two’s afternoon started with a paper by Dan Pope. Dan talked about a recent project concerning wind turbines, where he was tasked with auralising these for use in a VR application. The client brief made clear that a cost-effective solution was required. The solution used ambisonics recordings in typical conditions (which can be difficult to accurately capture). The resulting auralisation was used as part of a public consultation surrounding the replacement of an existing turbine with one with a quieter motor. Overall, the project was successful and the client was happy.

 

A subjective comparison of ambisonics rendered using virtual stereo microphone techniques
Bruce Wiggins from the University of Derby presented the second paper of the session which looked into different virtual stereo microphone techniques within an ambisonics system. This work was prompted by demand from the industry, where it was requested to decode to virtual stereo omni microphones. A series of tests, primarily relying on a multiple stimuli with hidden reference and anchor (MUSHRA) test, were used to judge the overall subjective quality of the resulting virtual microphone configurations. As could be expected, the results indicated that ‘it depends’ in terms of which stereo technique is best, but overall, it was proven that decoding to any arbitrary stereo microphone technique is possible within ambisonics.

 

Radial filter design for open spherical microphone arrays
The final paper of this session was presented by Nara Hahn (ISVR). Nara’s work focuses on the use and optimisation of open space spherical arrays, as opposed to the more traditional rigid spherical arrays. The aims of the array optimisation were efficiency, accuracy and robustness, where there would be one IIR filter for each notch. Additionally, the magnitude response of each filter was limited to less than 40 dB to avoid excessive peaks. The array was tested virtually with time, frequency and spatial performance was evaluated for a beamforming application. The results appeared to be promising, with further work identified.

 

SESSION EIGHT – Diversity presentation and discussion (Chair, Mark Bailey)
Mark Bailey introduced the diversity session at the conference by noting that the conference is still falling short on diversity but there are signs of encouraging progress. Mark reminded delegates that the number one rule for all of this is to be kind. Mark handed over to James Hipperson (Funktion One), who gave a thought-provoking and engaging talk on neurodivergence. He noted that neurodivergence is quite often overlooked but should really be celebrated as it provides diversity of thought within groups. A recent study by the Institution of Engineering and Technology (IET) found that one in five of its members identified as neurodivergent, suggesting that it is much more common in the RS (and wider technical) community than in other areas. As with all items related to diversity and inclusion, it is important to adjust, as appropriate, to improve things for everyone. A lively discussion followed James’s talk.

 

 

Above: Conference dinner at Mud Dock Café

 

SESSION NINE – Signal processing (Chair, Ludo Ausiello)
Complex-valued neural networks for the reproduction of single frequency sound fields
The final paper session of RS started with a paper presentation from Issac Lambert (ISVR), which looked into the issue of spatial aliasing due to microphone grid spacing within wavefield synthesis systems. A convolutional neural network (CNN) was used to help in this area. In this work, the target sound field was compared to the results from both a multilayer perceptron (MLP) and least square error (LSE). All were in agreement at low frequencies, but there were noticeable differences at higher frequencies. Nonetheless, the approach was shown to be effective for single frequencies, but further work is required to inspect it under more complicated scenarios.

 

Sound field analysis using a complex-valued neural network
The final paper of the conference was delivered by Vlad Paul (ISVR). In Vlad’s work, the focus was similar to that of the previous presentation but focused on the reconstruction of a single plane wave. Again, a complex-valued multilayer perceptron (MLP) was used. Even under test conditions with increased aliasing and noise, the results were quite good.

 

Conference close
Reproduced Sound 2024 was formally closed by Ludo Ausiello, who expressed his gratitude to all the presenters, session chairs, committee members and, of course, to the IOA’s Linda Canty for her years of service to the conference, wishing her a happy retirement. The papers were interesting and informative, resulting in a wonderful conference. He specifically thanked the technical crew from d&b audiotechnik and dBS, for their hard work throughout the event. We hope to see you all again in 2025 – bring your friends!

 

Below: Special presentation to Linda Canty on her retirement from the IOA