Loading...
Please wait, while we are loading the content...
Similar Documents
The Segregation of Frequency-modulated Concurrent Harmonic Sounds
| Content Provider | Semantic Scholar |
|---|---|
| Author | Cheveigné, Alain De |
| Copyright Year | 1996 |
| Abstract | Subjects were presented with pairs of concurrent vowels, and were requested to report one or two vowels for each stimulus. Vowels within a pair had either the same average fundamental frequency (F0), or F0s different by 3 or 6%. The F0 of each vowel was either constant, or else frequency-modulated (FM) at a rate of 4Hz with a peak excursion of 3%. Vowels were either both harmonic or both inharmonic (partials randomly displaced by -3%, 0% or 3%). One vowel within each pair was always stronger than the other by 15 dB, to reduce ceiling effects for the weaker vowel ("target"). The proportion of trials for which the weaker vowel was identified was recorded, together with the proportion of two-vowel responses. The experiment was expected to be sensitive to several mechanisms by which the identification of a target vowel might be affected by frequency modulation of that vowel, or of the competing vowel. The two-vowel response and identification rates were greater when the average F0s differed by 3% or 6% than 0%. Neither measure was affected by the presence of FM within the stimulus, or the modulation of the target or competing vowel. Identification was slightly less good when both vowels were modulated coherently (or static) than when modulation was incoherent. However incoherent modulation also implied greater maximum ∆F0, and one cannot rule out the possibility that this caused better identification. Overall, results agree with previous studies that suggest that FM has little or no effects beyond those attributed to FM-induced differences in fundamental frequency. A. Introduction Fourier analysis allows harmonic sounds, such as voiced speech, to be represented as a number of discrete components equally spaced in frequency. This has led to two expectations concerning the way such sounds are perceived in the presence of competing sounds. 1) The regular pattern of components, similar to the visible texture of a surface, should contribute to the perception of a harmonic sound that is partially occluded by interference. Components that share the regularity belong to the target, and y remaining components may be attributed to interference. 2) When the fundamental frequency of the sound changes, all its components shift in parallel on a logarithmic scale. This common movement should further aid the auditory system to group together components belonging to the target, and exclude components of other sounds that don't follow the same movement. Differently modulated sounds should be easy to segregate and perceive as separate entities. This idea, originally put forward by Helmoltz (1885), has become especially appealing within the framework of Auditory Scene Analysis (Bregman, 1990), as a prime example of grouping by "common fate". |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://audition-backend.ens.fr/adc/pdf/1996_ASA_Hawaii.pdf |
| Alternate Webpage(s) | http://www.ircam.fr/equipes/pcm/cheveign/sh/ps/hawaii96.pdf |
| Alternate Webpage(s) | http://www.ircam.fr/equipes/perception/cheveign/sh/ps/hawaii96.pdf |
| Language | English |
| Access Restriction | Open |
| Subject Keyword | Answer to reset Auditory system Bregman divergence Electronic component Entity Experiment FM broadcasting Fourier analysis Hearing Loss, High-Frequency Hideki Imai Interference (communication) Linear algebra Link's Awakening McGurk effect Modulation Obstruction Randomness Request - action |
| Content Type | Text |
| Resource Type | Article |
