Hello there!

To begin with, here are the links to:

The rest of this page contains audio samples of the concurrent speech configurations used in this project. These are best heard using stereo headphones or earphones.

Audio Clips:

The 5 Configurations

A sample audio file for each of the 5 configurations of variables can be found below. These are stereo recordings of the spatial audio file generated using Resonance Audio and Unity.

  • One Speaker
    • Audio Clip:
    • Headline: 25000 personnel to watch over election counting in Maharashtra
  • Two Speaker Same Voice
    • Audio Clip
    • Headline 1: 25000 personnel to watch over election counting in Maharashtra
    • Headline 2: Will make efforts to simplify GST to improve biz ranking: Finance Minister
  • Two Speaker Different Voice

    • Audio Clip

    • Headline 1: 25000 personnel to watch over election counting in Maharashtra
    • Headline 2: Will make efforts to simplify GST to improve biz ranking: Finance Minister
  • Three Speaker Same Voice

    • Audio Clip

    • Headline 1: 25000 personnel to watch over election counting in Maharashtra
    • Headline 2: Will make efforts to simplify GST to improve biz ranking: Finance Minister
    • Headline 3: CPI-M to launch 100 years celebration programme on Thursday
  • Three Speaker Different Voice

    • Audio Clip

    • Headline 1: 25000 personnel to watch over election counting in Maharashtra
    • Headline 2: Will make efforts to simplify GST to improve biz ranking: Finance Minister
    • Headline 3: CPI-M to launch 100 years celebration programme on Thursday

The Task Prompt

Given below is an example task prompt. You can then listen to the audio files sequentially to simulate the process of search for the relevant headline in pairs.

“You regularly follow Cricket, and often look for articles related to the Indian Cricket team whenever you read news headlines. In the following list of headlines, find the one that is most related to Indian Cricket.”

  • Items 1 & 2

  • Items 3 & 4

  • Items 5 & 6

  • Items 7 & 8

  • Items 9 & 10

  • Items 11 & 12

Concurrent Speech

A base Indian Accent voice from Amazon Polly was initially chosen for further processing.

  • Sample file of the base Indian accent female voice (Aditi)

Voice Synthesis using the Praat Software resulted in the following voice types:

  • Androgynous Voice

  • Male Voice

  • Female Voice

For the final experiment, in addition to the base voice, two other American accent voices were used.

  • Sample file of the American accent male voice (Matthew, Amazon Polly)

  • Sample file of the American accent female voice (Salli, Amazon Polly)

Thank you!