This page is the demo of audio samples for our paper. Note that we downsample the LJSpeech to 16k in this work for simplicity.

Part I: Speech Reconstruction

Recording		GT Mel + HifiGAN
GT VQ&pros + HifiGAN		GT VQ&pros + vec2wav
Recording		GT Mel + HifiGAN
GT VQ&pros + HifiGAN		GT VQ&pros + vec2wav
Recording		GT Mel + HifiGAN
GT VQ&pros + HifiGAN		GT VQ&pros + vec2wav
Recording		GT Mel + HifiGAN
GT VQ&pros + HifiGAN		GT VQ&pros + vec2wav
Recording		GT Mel + HifiGAN
GT VQ&pros + HifiGAN		GT VQ&pros + vec2wav

Part II: Text-to-speech Synthesis

with the particular purposes of the agency involved. the commission recognizes that this is a controversial area
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
this matter is obviously beyond the jurisdiction of the commission,
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
the commission regards this as a most useful innovation and urges that the practice be continued.
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
this no longer appears to be the case.
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
the secret service has been receiving full cooperation in scientific research and technological development
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
even if the manpower and technological resources of the secret service are adequately augmented,
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS
the commission recommends that the present arrangements
Recording	Tacotron 2	FastSpeech 2
GlowTTS	VITS	VQTTS

Part III: Prosody diversity in PL prosody hypotheses

county, and state law enforcement agencies in their districts.
Prosody 1	Prosody 2	Prosody 3
secret service personnel and facilities
Prosody 1	Prosody 2	Prosody 3
the assistance of trained federal law enforcement officers.
Prosody 1	Prosody 2	Prosody 3