AES Journal

Journal of the AES

2006 July/August - Volume 54 Number 7/8


A Robust and Computationally Efficient Speech/Music Discriminator (PDF-594KB) (HI-RES PDF-10.4MB)  
Jayme Garcia Arnal Barbedo and Amauri Lopes    571
Discriminating between speech and music is critical in such applications as speech recognition and automatic music transcription. A new method for achieving discrimination, which is computationally simple, attains almost perfect performance on a wide range of samples. The strategy uses four features extracted from frame histograms: zero-crossing rate, spectral rolloff, perceived loudness, and fundamental frequencies. Modest latency allows the strategy to be used in real-time applications.  

  Be the first to discuss this paper

Vibrations of Loudspeaker Cones in the Transitional Range (PDF-473KB) (HI-RES PDF-6.4MB)  
Zhi-Liang Zhang and Chang-Jun Cheng    589
Loudspeaker diaphragm vibrations are analyzed in the transitional frequency region where bending waves are beginning to play a significant role in relation to the longitudinal motion. An analytic and numeric analysis shows that the two types of motion are highly coupled. The first nonbending frequency can be considered to be the upper limit of the loudspeaker's frequency response.  

  Be the first to discuss this paper

Real-Time Processing of Image Sources Using Binary Space Partitioning (PDF-453KB) (HI-RES PDF-3.8MB)  
Dirk Schr��der and Tobias Lentz    604
Unlike the acoustic simulation of virtual spaces for static environments, with fixed source and listener locations, interactive virtual reality requires the aural experience to match a dynamically changing visual scene. Rapid computations are therefore more important than detailed acoustic modeling of a space. In the approach the geometry of a scene is subdivided into small subdivisions and pre-encoded in optimized data structures, BSP trees, with image source data.  

  Be the first to discuss this paper

Cumulative Spectral Analysis for Transient Decaying Signals in a Transmission System Including a Feedback Loop (PDF-652KB) (HI-RES PDF-19.0MB)  
Yoshinori Takahashi, Mikio Tohyama, and Yoshio Yamasaki    620
Determining the principle resonant frequency of a public address system before the beginning of howling is a valuable tool. By introducing a spectral accumulation function into the previously known technique of cumulative spectral analysis (used with loudspeakers), potential resonant frequencies are more clearly revealed even when they are only barely audible. When applied to the decaying portion of a speech signal, the display clearly predicts the principle resonance. Effectively, the technique increases the order of resonant poles.  

  Be the first to discuss this report

AES Standards Committee News (PDF-167KB)    630
True peak metering; digital interface for microphones; ATM network over Ethernet; IEEE 1394 use cases; digital audio measurements; digital audio input-output interfaces; audio-file transfer and exchange; transfer technologies  
120th Convention Report, Paris (PDF-6.1MB)    642
      Exhibitors (PDF-309KB)    660
      Program (PDF-1.0MB)    663
121st Convention Preview, San Francisco (PDF-286KB)    742
      Exhibit Previews (PDF-965KB)    744
Digital Radio Broadcasting (PDF-196KB) (HI-RES PDF-361KB)    771
Reviews of Acoustical Patents (PDF-1003KB)    636
News of the Sections (PDF-226KB)    775
Sound Track (PDF-130KB)    781
Available Literature (PDF-130KB)    782
Upcoming Meetings (PDF-117KB)    784
Membership Information (PDF-158KB)    785
Advertiser Internet Directory (PDF-161KB)    787
Sections Contacts Directory (PDF-193KB)    792
AES Conventions and Conferences (PDF-228KB)    800
Cover & Sustaining Members List (PDF-85KB)    
AES Officers, Committees, Offices & Journal Staff (PDF-152KB)    
Insert: 120th Convention Papers Order Form (PDF-119KB)    
AES - Audio Engineering Society