This supplementary material includes syncllm_samples.html file
containing audio samples. To hear the samples, you can open
the .html file in a web browser. The web browser will display
a table of stereo audio files with the following columns:

Prompt:      Spoken dialogue prompt audio
GT:          Ground-truth
Resynth-GT:  Ground-truth tokenized and resynthesized
             using the vocoder
dGSLM:       Dialogue continuation by dGSLM trained on Fisher
	     dataset
SyncLLM-F:   Dialogue continuation by SyncLLM trained on
	     Fisher dataset
SyncLLM-F-F: Full-duplex interaction between SyncLLM trained
	     on Fisher and SyncLLM trained on Fisher
SyncLLM-F-C: Full-duplex interaction between SyncLLM trained
             on Fisher and SyncLLM trained on Candor

The table contains 6 rows, in which first three rows contain
prompts sampled from the Fisher testset, and last three rows
contain prompts sampled from the Candor testset.

Note: Each audio file contains speech from two parties in the
dialogue recorded on separate audio channels. Use of headphones
is recommended to clearly notice individual streams generated
by different models.

