搜索结果: 1-15 共查到“知识库 语音处理”相关记录400条 . 查询时间(1.584 秒)
基于改进主题分布特征的神经网络语言模型
语音识别 语言模型 隐含狄利克雷分布 长短时记忆
2017/3/17
在递归神经网络(RNN)语言模型输入中增加表示当前词所对应主题的特征向量是一种有效利用长时间跨度历史信息的方法。由于在不同文档中各主题的概率分布通常差别很大，该文提出一种使用文档主题概率改进当前词主题特征的方法，并将改进后的特征应用于基于长短时记忆(LSTM)单元的递归神经网络语言模型中。实验表明，在PTB数据集上该文提出的方法使语言模型的困惑度相对于基线系统下降11.8%。在SWBD数据集多候选...
语音信号的混沌遮掩及其正定盲提取算法
小波变换 语音信号 FastICA算法 混沌系统
2016/9/1
利用小波变换提取语音信号的能量聚集带，将其隐藏在混沌载体信号中进行传输，设计一种盲提取算法实现不同混沌动力学系统下语音信号的有效提取。以3种不同维数的混沌动力学系统为背景，仿真实验定性和定量地分析了所提出算法的性能，验证了噪声环境下算法的可靠性，证明盲提取算法可作为对混沌保密通信系统保密性验证的有效方法。
Content-Based Tools for Editing Audio Stories
Audio editing storytelling transcript-based editing music browsing music retargeting
2016/5/24
Audio stories are an engaging form of communication that
combine speech and music into compelling narratives. Existing
audio editing tools force story producers to manipulate
speech and music track...
Audio producers often use musical underlays to emphasize
key moments in spoken content and give listeners time to re-
flect on what was said. Yet, creating such underlays is timeconsuming
as produc...
Capture-Time Feedback for Recording Scripted Narration
Voiceover narration speech emphasis audio
2016/5/24
Well-performed audio narrations are a hallmark of captivating
podcasts, explainer videos, radio stories, and movie trailers.
To record these narrations, professional voiceover actors
follow guideli...
Generating Emotionally Relevant Musical Scores for Audio Stories
Audio stories storytelling musical scores music retargeting
2016/5/24
Highly-produced audio stories often include musical scores
that reflect the emotions of the speech. Yet, creating effective
musical scores requires deep expertise in sound production
and is time-co...
FrameBase: Representing N-ary Relations using Semantic Frames
FrameBase Representing N-ary Relations Semantic Frames
2016/1/22
Large-scale knowledge graphs such as those in the Linked Data cloud are typically represented as subject-predicate-object triples. However, many facts about the world involve more than two entities. W...
约束条件下的结构化高斯混合模型及非平行语料语音转换
语音转换 结构化高斯混合模型 非平行语料 约束条件
2015/2/8
提出一种约束条件下的结构化高斯混合模型及非平行语料语音转换方法.从源与目标说话人的原始非平行语料中提取出少量相同音节，在结构化高斯混合模型的训练过程中，利用这些相同音节包含的语义信息及声学特征对应关系对K均值聚类中心进行约束，并在（Expectation Maximum，EM）迭代过程中对语音帧属于模型分量的后验概率进行修正，得到基于约束的结构化高斯混合模型（Structured Gaussian...
A Real Time Indoor Navigation and Monitoring System for Firefighters and Visually Impaired
Android Augmented Reality Spatial Database Visually Impaired OpenGL Firefighters
2014/12/8
There has been a widespread growth of technology in almost every facet of day to day life. But there are still important application areas in which technology advancements have not been implemented in...
Integrated CMOS IQ Upconverter/Downconverter for an X-Band Phased-Array Radar Application
RF CMOS X-Band Quadrature Conversion Injection Locking Quadrature Coupling Inegrated Mixer
2014/12/8
This thesis describes the design and measurement of an X-band IQ up/down converter that has been fabricated on a 180nm RF CMOS process. This converter includes components for mixing, frequency doublin...
基于最小控制GARCH模型的噪声估计算法
噪声估计 GARCH模型 MCRA算法 语音增强
2014/10/8
MCRA(Minima-Controlled Recursive Averaging)方法是经典的噪声估计算法,然而在语音段MCRA方法存在不能对噪声功率谱进行有效更新的问题.针对这一问题,本文利用广义自回归条件异方差(Generalized Autoregressive Conditional Heteroskedasticity,GARCH)模型在时频域对噪声信号建模,在MCRA算法原理的基础...
物理层安全的信号子空间人工噪声跳空方法
物理层安全 跳空 人工噪声 信号子空间
2014/2/25
现有人工噪声方法发送保密信号时在合法信道零空间内加入人工噪声以降低被截获概率，但当窃听者天线数不少于发送者时，该方法失效，针对这一问题，文中提出一种物理层安全的信号子空间人工噪声跳空方法。发送者随机选择信号子空间传输保密信号，并叠加人工噪声，合法接收者根据跳空图样选择相应信号子空间接收保密信号，并利用收发双方约定的噪声信息去除其中叠加的人工噪声，解调出保密信息，窃听者未知跳空图样和噪声信息，无法找...
为有效解决现有单一模型编码器无法在中低速率对语音和音频信号进行高质量通用编码的问题，本文借助语音与音频信号的谐波特性，建立了一种对语音和音频信号统一编码的方法。首先，本文利用经验模态分解(Empirical Mode Decomposition, EMD)提取输入信号的谐波成分；其次，利用感知匹配追踪算法，并结合正弦参数建模对谐波成分进行参数提取与量化；第三，对于量化谐波后的残差进行抖动格型矢量量...