The intelligibility of speech in rooms can be described by standardized objective measures (i.e. measurable quantities such as STI or others) or by subjective scores obtained in listening tests by means of a given test corpus (rhymed words, sentences etc..). Regarding the objective indicators, and STI in particular, there has been concern on how to deal with types of nonlinear processing such as compression or spectral subtraction, and also on how the peculiar effects of different kinds of time variant noises can be accounted for. This latter point is considered in the work by revising a former approach based on the wide band analysis of the noise fluctuation characteristics. The improvement consists in introducing an indirect estimate of the peculiar noise fluctuation impact. This approach is further developed by correcting for the different loudness values of noise and signal. The new objective indicator is matched with both the word intelligibility scores and the listening efficiency data. Two case history applications are reported to support the combined metric.
Revising fluctuation noise characteristics for describing the reception of speech in rooms with a combined metric
PRODI, Nicola;VISENTIN, Chiara
2013
Abstract
The intelligibility of speech in rooms can be described by standardized objective measures (i.e. measurable quantities such as STI or others) or by subjective scores obtained in listening tests by means of a given test corpus (rhymed words, sentences etc..). Regarding the objective indicators, and STI in particular, there has been concern on how to deal with types of nonlinear processing such as compression or spectral subtraction, and also on how the peculiar effects of different kinds of time variant noises can be accounted for. This latter point is considered in the work by revising a former approach based on the wide band analysis of the noise fluctuation characteristics. The improvement consists in introducing an indirect estimate of the peculiar noise fluctuation impact. This approach is further developed by correcting for the different loudness values of noise and signal. The new objective indicator is matched with both the word intelligibility scores and the listening efficiency data. Two case history applications are reported to support the combined metric.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
