Rabenfeder schrieb:Sehe ich auch so, aber wenn man sich die Videos von gestern ansieht
behauptet er bei 17.00 dass man aus dem geschlossenen Klo auf 177m
vermutlich nichts hören kann
Hier kann man das genau ausrechnen – hab mich jetzt damit mal näher beschäftigt (man hat ja sonst nichts zu tun;))
Wir wissen nicht, ob das Fenster offen war und Lin konnte es auch nicht ausschliessen. Für mich sind alle Hörzeugen glaubwürdig.
Zu deiner Rechnung, was für mich alles fremde Dörfer sind:
Rabenfeder schrieb:wenn man dann 20dB für den geschlossenen Raum abzieht bleiben noch immer 59,02 dB, ein Wert der gut zu hören ist.
Wenn Reeva um Hilfe geschrien hat, ich glaube, das war die Aussage Johnsons, dann wird sie das mMn wohl kaum aus einem geschlossen Fenster getan haben. Möglicherweise hat es Oscar dann später geschlossen.
Und noch ein Experte meldet sich dazu. Vielleicht könnt ihr ja mehr damit anfangen.
Hi there, I have a degree in Acoustical Engineering, I worked as a recording engineer for many years and now am a mastering engineer. I would say my expertise in the theory and practice of this field is easily a match for Mr Lin and although I can't promise to have all the answers, I am happy to answer and discuss anything websleuthers wish to ask.
I haven't caught up on today's testimony, but I heard yesterdays. Some obvious points struck me straight away:
The level of a scream at ~1m was assumed by Mr Lin to be 110dB, without specific citation or independent testing. However, research shows young, healthy females can easily be capable of hitting over 120dB at 1m when replicating a loud scream. You need to add 10dB to all his calculations straight off the bat. Please see the research paper here: http://www.audioforensics.com/PDFs/AES122_Scream.pdf (This is very loud, but it is nowhere near the same level as a jet at 1 metre - if Mr Lin said that, or implied 120dB at 1m from screaming would not realistically be possible, then I question his expertise)
The attenuation calculation in free air due to distance was fine
The attenuation due to the toilet door and open window at Pistorius' house was not a measurement - it was a modelled approximation - and I think Lin put it at about 29/30 dB. The range could be wide, but this is certainly on the high side - for location recordings, I would always assume 10dB indoors with window open, 20dB with a tight double-glazed window closed, and a further 10dB for each internal room/door division - this was important when minimising recording noise and in practice these measurements usually held up very well. So I would say 20dB not 30dB attenuation would be a more typical figure.
I can't remember off the top of my head the attenuation of Lin for a closed window at the receive location, but I remember thinking that too was an over-generous number - as above, I would model 20dB.
*Lin used the following criterion for audibility and intelligibility: signal must be greater than 10dBA above background noise level in dBA. I absolutely dispute this, it is seriously flawed and an over-stringent requirement in this specific scenario, for several reasons. I'll keep it brief for now but basically: 1) an A-weighting does NOT properly account for audibility across the whole frequency spectrum and all sound levels. Background noise at night is very-low frequency and low intensity, it is not very audible at all - a voice at the same dBA SPL will sound much louder - the A-weighting correction helps but doesn't go far enough - it simply does not work as Lin want's it to work, he is applying it inappropriately and far too simply. 2) Sounds can be audible even with negative signal-to-noise levels, especially when they occupy different frequency spectra 3) the character of a sound influences audibility to a significant degree - a scream in particular is known to be one of the most sensitive sounds, possibly through evolution - the criterion ignores this factor completely also 4) the paper I link to earlier simply disproves Mr Lin's criterion - there screams with a signal to noise level in dBA of just 4 and 5 dB were "clearly audible" and identifiable.
Mr Roux kept trying to do a cheap trick with the SPL meter and Mr Lin. He compared the SPL of the quiet courtroom (38dBA) - and said - see how quiet it sounds, this is what 38dBA sounds like. As I point out above, this is bogus - 38dBA of talking for example would still be quiet but it would be perceived as much louder than the background noise of a room.
*If I have time tonight, I will do a quick bit of research - at my home, I will measure the background noise level and see exactly what level of scream is needed to be audible and identifiable. This will serve as a far more accurate model of what SPL is typically needed (IMO) than Mr Lin's criterion.