Slide 1

Slide 1 text

LINE AI LINE AI Voice / Voice

Slide 2

Slide 2 text

/ Ryuichi Yamamoto 1 › Engineer/researcher @ Voice team › Joined LINE in 2018 › Work – GPU text-to-speech – Neural vocoders – Controllable TTS Test – is

Slide 3

Slide 3 text

LINE 2

Slide 4

Slide 4 text

Voice [ ] Parallel WaveGAN: GPU LINE DEV DAY2020 – https://linedevday.linecorp.com/2020/ja/sessions/8721 [ ] NLP Engineering Blog – https://engineering.linecorp.com/ja/blog/newgrads-nlp-text-to-speech/ [ ] AI Park Byeongseon | 2023 – https://linecorp.com/ja/career/newgrads/interview/2022/engineering08 3

Slide 5

Slide 5 text

4 LINE

Slide 6

Slide 6 text

Voice (2022/04/26 ) – ICASSP 2021 2 – ICASSP 2020 4 – Interspeech 2021 2 – Interspeech 2020 1 – Interspeech 2019 1 – SLT 2021 1 – ASRU 2019 1 – ASJ 2021a 1 – ASJ 2020s 1 – 2021 1 – NL/SLP 2021 1 5 2 6 8 1 2 3 4 5 6 7 8 9 2019 2020 2021

Slide 7

Slide 7 text

Light-weight Controllability Topics Low-resource https://linedevday.linecorp.com/2020/ja/sessions/8385 ここまで来た音声技術・今後の展望 (LINE DEVELOPER DAY2020)

Slide 8

Slide 8 text

Voice – – / 1 – – J – – Wav2vec self-supervised learning – – – / – 7