Clear and Casual speech differences

Outside noise

• Clear sentence
• Casual sentence

You will hear 8 sentences inside -2dB and -8dBSpeech Shaped Noise (SSN). The original speech sentence is modified with three algorithms. The Spectral Shaping and Dynamic Range Compression algorithm (SSDRC) has been proposed by Catalin Zorilla and Yannis Stylianou and has been shown to be high intelligible in noise. The Mix-filtering approach proposed by Maria Koutsogiannaki has been tested in SSN and reverberant conditions and has been found to increase speech intelligibility while maintaining the quality of speech. Mix-filtering modifies the spectrum of speech similar to that of Clear speech. The last proposed scheme DMod enhances the depth of the temporal envelope modulations and applies a non-linear compression function to the temporal envelopes similar to what the ear does. All signals have the same RMS energy Quality of the Mix-filtering

• Mix-filtering
• Lombard inspired

Also you will hear signals in reverberation.

• -2 dB SSN
• Original sentence
• SSDRC
• Mix-filtering
• DMod
• -8dB SSN
• Original sentence
• SSDRC
• Mix-filtering
• DMod

Greek sentences in SSN -8dB SNR

• Original sentence
• SSDRC
• Mix-filtering
• DMod

English sentences in reverberation with RT2=2sec

• Casual sentence
• Clear sentence
• Mix-filtering
• Uniform time-scaling
• Mix-filtering and Uniform time-scaling

Refresh page to hear same sentence twice