Audio Samples for "DSST: Wireless Deep Speech Semantic Transmission"

Channel Bandwidth = 10KHz, SNR = -1dB, AWGN Channel

Audio 1

Audio 2

Audio 3

Audio 4

Original audio

DSST (Our model)

AMR-WB + 5G LDPC

Opus + 5G LDPC

DeepSC-S

(Bandwidth = 12KHz)

Channel Bandwidth = 4KHz, SNR = 5dB, AWGN Channel

Audio 1

Audio 2

Audio 3

Audio 4

Original audio

DSST (Our model)

AMR-WB + 5G LDPC

Opus + 5G LDPC

Channel Bandwidth = 10KHz, SNR = 2dB, COST2100 Fading Channel

Audio 1

Audio 2

Audio 3

Audio 4

Original audio

DSST (Our model)

AMR-WB + 5G LDPC

Opus + 5G LDPC

DeepSC-S

(Bandwidth = 12KHz)