Baidu Online Network Technology (beijing) Co., Ltd.

Beijing Didi Infinity Technology and Development Co., Ltd.

Baidu Online Network Technology Co., Ltd.

Xiulin Li

Jie Bai

Yan Xie

Fengmin Gong

Qingchang Hao

Haiyuan Tang

Systems and methods are provided for language-based service hailing. Such system may comprise one or more processors and a memory storing instructions that, when executed by the one or more processors, cause the computing system to obtain a plurality of speech samples, each speech sample comprising one or more words spoken in a language, train a neural network model with the speech samples to obtain a trained model for determining languages of speeches, obtain a voice input, identify at least one language corresponding to the voice based at least on applying the trained model to the voice input, and communicate a message in the identified language.

System and method for language-based service hailing

Systems and methods for speech recognition are provided. The method may include obtaining a plurality of candidate recognition results of speech information uttered by a user and a plurality of preliminary scores corresponding to the plurality of candidate recognition results, respectively. The method may further include, for each of the plurality of candidate recognition results, extracting one or more keywords from the candidate recognition result and determining at least one parameter associated with the one or more extracted keywords. The method may further include, for each of the plurality of candidate recognition results, generating an updating coefficient based on the at least one parameter and updating the preliminary score based on the updating coefficient to generate an updated score. The method may further include determining, from the plurality of candidate recognition results, a target recognition result based on the plurality of updated scores.

Systems and methods for speech recognition

Embodiments of the disclosure provide systems and methods for segmenting a sentence. The method may include identifying a first phrase in the sentence associated with a first segmentation path, determining a first group of derivative phrases semantically associated with the first phrase, determining a first evaluation score based on modified sentences generated by replacing the first phrase with the respective derivative phrase in the first group, and segmenting the sentence based on the first evaluation score.

System and method for segmenting a sentence

An acoustic model generation method and device, and a speech synthesis method and device. The acoustic model generation method comprises: acquiring personalized data, wherein the personalized data is obtained after processing according to personalized speech data and corresponding personalized text data (S); acquiring a pre-generated reference acoustic model, wherein the reference acoustic model is generated according to existing large-scale samples (S); and carrying out self-adaptive model training according to the personalized data and the reference acoustic model to generate a personalized acoustic model (S). According to the method, an acoustic model can be rapidly generated, and personalized requirements of users can be satisfied.

Acoustic model generation method and device, and speech synthesis method

A training method for multiple personalized acoustic models, and a voice synthesis method and device, for voice synthesis. The method comprises: training a reference acoustic model, based on first acoustic feature data of training voice data and first text annotation data corresponding to the training voice data (S); acquiring voice data of a target user (S); training a first target user acoustic model according to the reference acoustic model and the voice data (S); generating second acoustic feature data of the first text annotation data, according to the first target user acoustic model and the first text annotation data (S); and training a second target user acoustic model, based on the first text annotation data and the second acoustic feature data (S).

Training method for multiple personalized acoustic models, and voice synthesis method and device

The present invention provides a method and a device for optimizing speech synthesis system. The method comprises: receiving speech synthesis requests contained text messages; and determining the load level of the speech synthesis system when the speech synthesis requests are received; and selecting speech synthesis paths corresponding to the load level and synthesizing the text into speech according to the speech synthesis paths.

Method and device for optimizing speech synthesis system

The present disclosure provides a speech synthesis method and apparatus. The speech synthesis method includes: processing a text, to obtain a to-be-synthesized text; if a network connection exists, sending the to-be-synthesized text to an online speech synthesis system for speech synthesis; and if a fault occurs in the online speech synthesis system in a process in which the online speech synthesis system performs speech synthesis or the network connection is disrupted in an actual use process, sending a text for which the online speech synthesis system has not completed speech synthesis to an offline speech synthesis system for speech synthesis.

Speech synthesis method and apparatus

The present invention discloses a method and apparatus for speech synthesis based on a large corpus. The method for speech synthesis based on a large corpus comprises: utilizing a prosodic structure prediction model to carry out prosodic structure prediction processing on input text to provide at least one alternative prosodic boundary partitioning solution; determining a prosodic boundary partitioning solution according to structure probability information about a prosodic unit in a speech corpus in the at least one alternative prosodic boundary partitioning solution; and carrying out speech synthesis according to the determined prosodic boundary partitioning solution. The method and apparatus for speech synthesis based on a large corpus provided by the embodiments of the present invention improve the naturalness and flexibility of speech synthesis.

Growing community of inventors

Beijing, China

Xiulin Li