A Research of Speech Signal of Fire Information Display Interface

Liang Zhang, Xianghong Sun, Thomas PlocherHCI 2007 · DOI

In high-stress fire emergencies, firefighters are often confronted with a "data deluge" on visual displays, ranging from building schematics to fireplug locations. This visual overloading can lead to cognitive processing delays and life-threatening errors. While auditory signals are processed faster by the human brain, their effectiveness is not universal; intelligibility depends heavily on language phonetics, background noise, and vocal characteristics. This research investigates how to optimize Chinese-language auditory fire alerts to maximize information transfer for emergency responders.

Experimental Methodology

The researchers employed a 4-factor, within-subjects experimental design involving 36 participants. The study manipulated four key variables to measure their impact on accuracy (correct repetition of phrases) and reaction time (RT):

  1. Speech Rate: Ranging from 5 to 9 characters per second.
  2. Vocal Tune: Soprano, Mezzo-soprano, Tenor, and Baritone.
  3. Speech Type: Nouns, Numerals, General sentences, and Numeral-heavy sentences.
  4. Noise Level: 0 dB (silent), 60 dB, and 80 dB (simulating high-noise emergency environments).

The Impact of Environmental Noise

Environmental noise is a primary barrier to communication in fire scenarios. The study confirmed that as decibel levels increase, speech intelligibility significantly degrades.

Impact of noise levels on speech intelligibility

As shown in the data, accuracy peaked at 88.1% in silent conditions but dropped to 65.7% at 80 dB. Interestingly, while accuracy decreased at the highest noise level, reaction times actually became shorter compared to the 60 dB condition. This suggests a "urgency paradox" where the high-stress nature of the loudest noise level prompts faster, albeit less accurate, responses.

Optimizing Speech Rate and Intelligibility

Finding the "sweet spot" for speech delivery is crucial for balancing speed and comprehension. The research indicates a clear inverse relationship between the rate of delivery and the accuracy of the listener.

Relationship between speech rate, accuracy, and reaction time

The study found that for general descriptive sentences, a rate of 7 characters per second is optimal. However, for more complex information, such as isolated nouns or sentences containing numbers, the rate must be slowed to 5–6 characters per second to maintain acceptable accuracy levels. Linguistic context helps the brain "fill in the gaps" caused by noise, allowing for faster processing of full sentences compared to isolated words.

Vocal Characteristics and Content

Beyond speed and noise, the vocal range of the alert plays a vital role. The mezzo-soprano (mid-high range female voice) proved to be the most robust against masking by background noise. In contrast, lower-pitched male voices, such as baritone, were much more likely to be obscured by environmental sounds.

Furthermore, the fragility of numerical data was highlighted: accuracy for numeral-heavy sentences dropped faster than for general sentences, requiring a more conservative delivery speed to ensure critical data like floor numbers or pressure readings are correctly understood.

Conclusion and Recommendations

To minimize cognitive load and maximize safety, the researchers recommend that Chinese fire information interfaces adopt the following parameters:

  • Optimal Tune: Mezzo-soprano.
  • Optimal Speech Rates: 5 characters/s for isolated words; 6 characters/s for numeral sentences; 7 characters/s for general descriptive sentences.
  • Contextual Delivery: Use full sentences where possible to leverage linguistic redundancy against noise.

Future research will involve professional firefighters to account for the "acquaintance effect" and explore the synergy between auditory signals and visual multi-modal interfaces.

Summary updated

BibTeX
@inproceedings{zhang2007research,
  author = {Liang Zhang and Xianghong Sun and Thomas Plocher},
  title = {{A Research of Speech Signal of Fire Information Display Interface}},
  booktitle = {HCI},
  year = {2007},
  doi = {10.1007/978-3-540-73331-7\_94},
}
Figure