Category Archives: Discussion

Gunshots, noise off – and music

When acosusticians investigate the acoustic response of rooms they normally excite the room in such a way that they are able to study the room’s acoustic response to a perfectly impulsive sound.  This response is called the room’s echogram or reflectogram. Based on this echogram a set of acoustic measures can be calculated, for instance the reverberation time. Calculating the reverberation time is based on observing the long decay after a long-lasting sound source is switced off. In the past balloons were popped off, gunshots were fired off and/or loudspeakers switched off to obtain the relevant responses. With modern measuring techniques the same responses are obtained by use of loudspeaker and taylor made excitation signals like sine sweep. Most of the time the loudspeaker as well as the microphone are omnidirectional since this makes it easier to obtain reliable results among different investigations and between acousticians.

On the other hand music rarely contains perfectly impulsive sounds or long-lasting constant sounds being switched off (the ‘noise off’ method). Apart from at very high frequencies musical instruments excite the room over a certain time, leading to reflections from the room fusing together. Addtionally, the instruments are highly directional and we hear sound in stereo, not mono. Perceptual effects like masking and the cocktail party effect appear relevant for a listener inside a reflective room. Accummulated early reflections as the music is running can contribute to mask the direct sound and make it difficult to separate different sound sources spatially. This can as proposed by David Griesinger make the sound appear more muddy and the listener will not feel engaged in the sound. Since localisation is affected by the shape of our head and outer ear, the extent to which a sound appears defined or fully meddled can only be judged by an individual listening inside the actual room.

Based on this we may claim that measurability and reliability has been given priority compared to validity, when assessing room acoustic spaces. Focusing mainly on measured characteristics may have limited the scope and approaches when searching for relations between architectural design and experienced conditions. My impressions based on the results from the PhD work is that acoustic measures are relevant to see if the proposed design is ‘in the ballpark’, to ensure that the most catastrophic results can be avoided – formally or just to ensure the end users are happy with the acoustics.

One paradox within room acoustics may be that even if absolute acoustic levels are widely used within other fields of acoustics, like formal limits for sound pressure levels from traffic noise, it is not (yet) an established single measure that is always included when assessing listener and performer acoustics. The acoustic measure Strength (G) represents the absolute acoustic gain by room, but is not always measured/reported. For stage conditions, absolute levels are measured, but by use of a different reference level (the Support ST measures).

In my view, the acoustic measures studied for performance spaces should at a minimum include T, EDT, G and C80, both in the audience area and on stage. For the audience area spatial measures are also relevan (like LFC and LEV). From measured G and C80, Gearly and Glate should also be found to provide an indication of how loud the reverberation is, not only the reverberation time (also see ‘How loud is my reverberation‘ by Griesinger).

In addition to obtaining these measures it appears highly beneficial to listen a lot to rooms with own ears, and as a substitute, listen to auralisations of potential responses for planned venues. The acoustic measures have not been developed much since the ’70s and to progress further I believe experiencing real spaces  and discussing them are very useful to test and develop new hypothesis regarding links between physical acoustic conditions and experiences.

By having a common understand of basic acoustic concepts among for instance acousticians, sound engineers and musicians it will be much easier to discuss and exchange ideas and experiences based on qualified listening in real rooms (with ensemble on stage, audience present, directivity taken into account etc etc). Such qualitative discussions may be a very fruitful complement to objective acoustic data. If musicians and sound engineer also can understand the essential use and relevance of acoustic data I believe we are heading towards exciting discussions and new exciting findings within the discipline of stage acoustics.

Get the basic measures right and be aware of their limitations.  Then make sure to have enough time and interaction to listen, discuss and explore acoustic spaces with open ears and an open mind! I believe this will make it easier for us (and more enjoyable as well?) to effectively create great acoustic spaces for everyone involved.

Why G appears superior to ST for acoustic levels

For more details and background on why measures based on G appears to be both more reliable and subjectively relevant compared to the ST measures, I have written a 3 pages long article that include some results previously not presented. See the Articles page.

The message from the article in brief: feel free to continue measuring ST on stage, but please make sure that you also measure and report T, C80, Ge and Gl on stage as well as in the audience area for the venues you investigate!

Creating a fuller sound of the orchestra

A full or warm sound of the orchestra is normally preferred. An experience of a full sound is normally associated with sufficient levels and acoustic response at low frequencies. An increase of reverberation time at low frequencies is often recommended for this purpose. By having bass-reflecting surfaces close to the orchestra at the sides and at the back, close to coherent reflections will be added to the source that will effectivily boost the sound source levels at low frequencies. The need for low frequency reverberant sound to produce a fuller orchestra sound may therefore be reduced with surfaces close to the orchestra. Such proximate bass-reflecting surfaces can therefore potentially contribute to both a full (warm) and distinct orchestra sound.

This apparent benefit of side and back reflecting surfaces was covered in my PhD thesis with emphasis on raising the levels of the double basses. But the percussion section may also benefit from being close to surfaces that reflect at low frequencies. The back wall has been associated with negative effects, like unnecessary raising the level of percussion. But such negative effects of percussion could to be relevant primarily at frequencies above 500 Hz. The back wall behind the percussion can be made reflective at low frequencies, while more absorbing or diffusing (sound scattering) at higher frequencies to avoid the negative effects. The brass instruments appears loud enough in the first place so there may be no need to raise the levels of brass at all with reflecting surfaces close to them. The results from the questionnaire studies suggest that the exception could be the horn section. Horn players have commented positively on having a reflecting surface behind them, that could be caused by the directivity of the horn.

Phase relations between sound waves are normally not included when sound levels are summed in computer simulating software for auditorium acoustics. By ignoring phase information the effect of the surfaces will be underestimated regarding total levels at low frequencies. For a source close to one reflecting surface, calculated acoustic gain G (Strength) from the simulation software can be typically 2–3 dB lower than the real value. If making design desicions based solely on the resulting G values from software the need for long reverberation time at low frequencies can be overestimated and resulting in too high levels in the bass for the orchestra and other users of the hall, like pop/rock bands

Our paper published in JASA (more or less Chapter 4 of my PhD thesis) demonstrates how to calculate the combined level based on the direct sound and a reflection with phase relation taken into account. A spreadsheet is also available for carrying out such calculations (comb filtering interference), see the Spreadsheets page. The total level presented in the spreadsheets will be relative to the direct sound level. If knowing the source-receiver distance, the total value of G can be found. If exporting wave files for auralisation from the computer simulation software, phase relations between the direct sound and early reflections can be included, but then there can often be a problem with obtaining correct values of G based on the exported (often normalised) wave files.

Why bother with calibrating for measuring G?

The acoustic measure G (Strength) represents the total level of the acoustic response, in other words how much an acoustic space gains the sound level within an acoustic space. The reference level for G is the direct sound level at 10 m from the  source. Measuring G requires that the measurement system is calibrated and this is probably the major reason for why this measure often is not included when acousticians investigate acoustic conditions. The calibration is normally done by testing the measuring equipment in use within an anechoic chamber, where the direct sound level at 10 m is obtained. It can alternatively be done in-situ (at the location), but this puts some restriction on the frequency range where correct values of G are obtained. A brief article on how to calibrate or confirm G in-situ (that will result in sufficiently valid values of G from the
250 Hz octave band and above, dependent on your source) is available on the Articles page.

So why bother with calibrating the system for being able to measure G? From measured G and C80 the level of the early and the late (reverberant) acoustic response from an acoustic environment can be obtained – namely Ge and Gl, the level G for the early (e) acoustic response arriving before 80 ms  and the late (l) acoustic response arriving after 80 ms (relative to the direct sound arrival). The reverberation time T and the level of the reverberation appears to be essential attributes of an acoustic space. By knowing the level of the reverberation (as well as the early acoustic response) we can tell for instance if an acoustic space will contribute to an excessively loud environment or if the reverberant response will be inaudible. If knowing the reverberation time T and the volume of the acoustic space V, the level of the acoustic response can be estimated. But the estimate can often be too far off from the actual situation, so by actually measuring the acoustic level we will get rid of some uncertainty factors.

Recent articles where the relevance and benefits of measuring G are discussed includes Barron (2008) , Buen (2010) and Beranek (2011), the latter is unfortunately not freely available (abstract only). Regarding the relevance of G and Gl for stage acoustics, see the articles on suggested objective assessment of concert hall stages, Dammerud et al (2010) and Dammerud (2011).

Also see the Spreadsheets page for spreadsheets on how to calculate Ge and Gl based on measured G and C80 or estimate Gl based on measured T and hall volume V.

On early reflections in concert halls

Lokki et al. have recently published an express letter in JASA proposes that large and flat surfaces surrounding the listeners are better than diffusing surfaces. It also suggests that reflecting surfaces above the listener and stage contribute to reduce the perceived sound quality. In concert halls early reflections are normally needed to raise the total perceived sound level. According to this paper, early reflections from directions outside the sagittal plane that have a frequency and phase response comparable to the direct sound are most effective for boosting loudness without leading to negative sound characteristics. This express letter is freely available and includes a video with auditory demonstrations.

This proposal support the findings from the research project presented in this website: that a narrow and high stage enclosure is more beneficial than a wide and low, and that the reflecting surfaces close to the orchestra at the sides can be made flat and non-diffusing to effectively compensate for low direct sound levels from strings.

David Griesinger has studied how early reflections and reflected sound in general affect our engagement as listeners. In a recent paper presented this year, available from his web site, he proposes a psycho-acoustic model that links aspects of perceived speech and music to the physical acoustic response from an enclosed space. The proposed model differ from the precedence effect, but implication of the model also agree with the conclusion that a high stage enclosure is beneficial for an orchestra.

Whereas the model of Lokki et al. favours  early reflections from the sides, Griesinger’s model doesn’t. But they both demonstrate that the acoustic measure C80 has limited relevance for perceived clarity of the sound. The last word has definitely not been said regarding early reflections within performance spaces.