ANALYSIS of MUSIC PERFORMED IN DIFFERENT ACOUSTIC SETTINGS in STAVANGER CONCERT HOUSE Tor Halmrast Statsbygg 1.ammanuensis UiO/Musikkvitenskap NAS 2016
SAME MUSIC PERFORMED IN DIFFERENT ACOUSTIC SETTINGS: FARTEIN VALEN KONSERTSAL 2.Chamber 3.Amplified 6. Concert Big(tilslutt) ZETLITZ (Flerformåls-sal) 5. Zetlitz
Espen Berens trpt. UiS Carl Haakon Waadeland drums, NTNU Jan Tro recordings, NTNU Svein Folkvord double bass/el-bass, Sinus Akustikk a/s String quartet from the 1B1 ensemble (UiS), Tor Halmrast guitar and piano
ZETLITZ multi- purpose hall
Surprisingly, the parameters most common for Musical Information Retrieval (MIR) etc. show NOT to be able to detect the very clearly perceived timbral differences between the settings. The reverberation times (and investigations of differences in tone length is the only way to show the differences. This seems to be a big hole in the repertoire of MIR analysis.
MUSIC
6. CONCERT BIG 5.ZETLITZ
STRENGHT
sec.
CALCULATED changes in G[dB] In a diffuse field: G = 10 log (RT/V) + 45 db For our medium volume (2.Chamber and 3.Amplified) changing RT between 2 to 1.7 s reduces G 1,6 db For a constant RT of 2 sec, changing the height 4-6 m gives a rise in G of 0.9-1,4 db
Lp COMPARED TO MEASURED G[dB] Free hanging side-balconies with downstands. A wanted effect of this is that: Kahle s Acoustics Measurements Report shows that the early strength G,early changes just 0.1 db dampened/fully reverberant The G, late, however, increases 1.6 db from 2.Chamber to 6.ConcertBig. Good agreement with the 0 db change found from the recordings. (0.0 to 2.0 db, with the highest values for the peak sounds ). The almost zero change of strength for the mean and Leq values might indicate a slight compensation from the musicians, meaning that they play somewhat softer in the most reverberant setting ( 6.ConcertBig ). (Earlier measurements in music rehearsal rooms show a bigger compensation, see Halmrast: Musician s Perceived Timbre and Strength in (too) Small Rooms ).
5.ZETLITZ
SPECTRAL CENTROID opposite trends. Recordings: Zetlitz includes less reflections/reverb, more direct sound, and shows a higher amount of treble. Reverberation time measurements: the big, non-dampened Valen shows the most brilliant reverberation time. (PS! The flexible absorbers in Valen are curtains, so the damping is mainly for mid-/high frequencies).
The Amount of reverberation might influence the reduction of brilliance more than the actual Rev. Time v/frequency. This might indicate that Adding reverb will reduce high frequencies (also if the reverb has an increase of reverberation for high frequencies). (This is of course somewhat analogue to how pass filters are made, taking the mean of several delayed samples, like a smoothing of the signal).
TONE LENGTH ATTACK MIR-attacklength: Does not find each note 17 khz 0 Hz
ZETLITZ 1.5 khz CONCERT BIG
1.8 khz STRING QUARTET
TRUMPET a capella Fanfare The very shortest tones in 5.Zetlitz seems longer than for the other settings. This is the trumpeter s way of compensating for the lack of reverb in Zetlitz.
4 khz JAZZ-PIANO-TRIO 300 Hz MIR does not manage to detect each piano lead attack
Tempo is highest in 5.Zetlitz (with the shortest reverberation time). 5.Zetlitz is ca. 5% faster than in 2.Chamber. Tempo is high also for 6.ConcertBig is surprising: last recording before going home, or that the faster tempo from Zetlitz was still in the blood of the musicians?
ENVELOPMENT Inter Aural Cross Correlation Increased envelopment (lower correlation) for 6.ConcertBig compared to the dampened Zetlitz and 2.Chamber. Measurements (Kahle Acoustics), Valen, main configuration : 1-IACC,0-80ms 0.65 dvs.: IACC, 0-80ms 0.35 Our stereo recordings of course give higher correlation than binaural measurements
PERCEIVED LOUDNESS grows with duration up to 200 ms
CONCLUSION THE INFLUENCE OF REVERB ON ATTACK/DECAY: Long reverberation in Valen attacks and decays overlap, Blurring of short, transient tones (jazz piano) Allows full development of long tones (strings) Reverb SMOOTHING Longer Attack-time and Decay(release) time, slight reduction of high frequencies Longer attack reduction of PERCEIVED treble Very short sounds longer duration=higher PERCEIVED loudness