公开(公告)号：US20200135172A1

公开(公告)日：2020-04-30

申请号：US16666043

申请日：2019-10-28

Applicant: Google LLC

Inventor： Yutian Chen ,  Scott Ellison Reed ,  Aaron Gerard Antonius van den Oord ,  Oriol Vinyals ,  Heiga Zen ,  Ioannis Alexandros Assael ,  Brendan Shillingford ,  Joao Ferdinando Gomes de Freitas

IPC: G10L13/047 ,  G06N3/08 ,  G10L13/033 ,  G10L13/04

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating an adaptive audio-generation model. One of the methods includes generating an adaptive audio-generation model including learning a plurality of embedding vectors and parameter values of a neural network using training data comprising first text and audio data representing a plurality of different individual speakers speaking portions of the first text, wherein the plurality of embedding vectors represent respective voice characteristics of the plurality of different individual speakers. The adaptive audio-generation model is adapted for a new individual speaker using adaptation data comprising second text and audio data representing the new individual speaker speaking portions of the second text, the new individual speaker being different from each of the plurality of individual speakers, wherein adapting the audio-generation model includes learning a new embedding vector for the new individual speaker.

公开(公告)号：US20230401451A1

公开(公告)日：2023-12-14

申请号：US18199886

申请日：2023-05-19

Applicant: Google LLC

Inventor： Yutian Chen ,  Xingyou Song ,  Chansoo Lee ,  Zi Wang ,  Qiuyi Zhang ,  David Martin Dohan ,  Sagi Perel ,  Joao Ferdinando Gomes de Freitas

IPC: G06N3/0985 ,  G06N3/0455

CPC classification number: G06N3/0985 ,  G06N3/0455

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model. One of the methods includes receiving metadata for the training, generating a metadata sequence that represents the metadata, at each of a plurality of iterations: generating one or more trials that each specify a respective value for each of a set of hyperparameters, comprising, for each trial: generating an input sequence for the iteration that comprises (i) the metadata sequence and (ii) for any earlier trials, a respective sequence that represents the respective values for the hyperparameters specified by the earlier trial and a measure of performance for the trial, and processing an input sequence for the trial that comprises the input sequence for the iteration using a sequence generation neural network to generate an output sequence that represents respective values for the hyperparameters.