ResearchLabPresentation
- 1. Can People Segregate Sounds in Two Different Rooms? Brianna Weibye University of Minnesota
- 2. Are We Sensitive to Room Acoustics? • Use room acoustics to detect reverberance We are sensitive to room acoustics and use information to organized auditory objects in distance. • Segregating the direct portion, not the reverberation tails (stays the same). Gives information about global environment. • Evidence of segregation of two objects based on tail of temporal envelope. Can be done if one of the objects has no tail (target or masker). TIME INTENSITY Direct Early Reflections Reverberation TIME INTENSITY Direct Early Reflections Reverberation Reverberation time (T60) = Time (in sec) for the intensity to decrease by 60 dB (a factor of 1,000,000).
- 3. Overview: Introduction Goal: Can people segregate sounds in two different rooms? Overall Concepts: Used two tasks -- 1) Detection – Can subjects detect which rhythm is more reverberant? - Used detection thresholds to see how sensitive our subjects were and how they can compare reverberant stimuli. 2) Discrimination – Can subjects segregate the rhythms from the maskers? - Can subjects discriminate objects over time using room acoustic cues?
- 4. Virtual Room Model How good are people detecting room acoustics based on t60? • Model we used based on Zahorik, 2009. How virtual environments are made? 1) Image models used for early reflections. This gets early portions and impulse responses. In our case 500 reflections are simulated First 500 reflections are spatialized, they sound like they’re coming from somewhere in space. After 500 it decays and becomes just noise. 2) Statistical Model simulates late reverberation models. Based on Sabine equation: relates volume of room to how quickly it decays. Based on absorption and reflection surfaces • From this, we can now create a virtual room using headphones. • Important: We did not change the dimensions of the room only energy absorbed. Timing of early reflections are exactly the same.
- 5. Methods of Detection Task White Noise Burst at 50ms Convolved this noise with impulse response (“Putting it in the room”) . Remove leveling as a que by level roving ( +/- 6dB). Tracking is 3 down-1 up adaptive track Step Sizes: (1000ms, 500ms, 60ms, 10ms) Steps decrease with downward reversal • 400ms quiet office • 1000ms auditorium
- 6. Results of Detection Task More variability in 1000ms case than 400ms Mean data significantly different from the reference. 400ms Mean Threshold: 1000ms 1000ms Mean Threshold: 2200ms “Just Noticeable Differences” were about the same (between 1.25-1.5)
- 7. Methods of Discrimination Task Same 50ms noise bursts. 4 repetitions of a rhythm (each trial is about 11 seconds). Practiced target rhythm only conditions until they reached 90% with no masker, usually took two or more tries. Practice in beginning of each trial before actual measurement. Random masker placements; each trial had random rhythm, and then asked which rhythm they heard. Parametrically varied masker t60, using two target references (400ms, 1000ms) 3 Conditions: 1) Monaural 2) Binaural 3) Binaural 90 degree ** Expected binaural 90 degree to be better based on acoustic head shadow providing a better ear advantage for reverberant listening. Performance on the rhythm task is an objective measure of segregation.
- 8. Results (400ms Reference) • Average mean across all subjects • Each condition represented by different color (monaural, binaural, binaural 90 degrees) • The target (grey) stayed the same. We changed the masker (duration)
- 9. Results (1000ms Reference) • Detection was off the measurements we tested • Similar results to the 400ms Reference
- 10. Conclusions • People can segregate anechoic from a reverberant environment but cannot discriminate the influence of two differing reverberant environments on auditory objects. • Most everyday environments are between 400ms (e.g small office) and 1000ms (e.g auditorium), so we are exposed to these types of environments all the time. • We can at best detect the difference in reverberance between a small office and an auditorium, however, we cannot use this information to discriminate which objects are in which room. • The brain might have difficulty tracking more than one environment at one time. Which makes sense because you’re only in one room at once.