中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

122 records (took 0.022 seconds)

    Speech analysis system
    1.
    发明授权
    Speech analysis system 失效
    语音分析系统

    公开(公告)号：US3479460A

    公开(公告)日：1969-11-18

    申请号：US3479460D

    申请日：1966-12-29

    Applicant: IBM

    Inventor： CLAPPER GENUNG L

    IPC: G10L19/02 ,  G10L1/02

    CPC classification number: G10L19/02

    Abstract: 1,154,129. Speech recognition. INTERNATIONAL BUSINESS MACHINES CORP. 20 Dec., 1967 [29 Dec., 1966], No. 57991/67. Heading G4R. In speech analysing apparatus, harmonics and the fundamental are detected, detection of the latter controlling means providing signals indicating the time of occurrence of each harmonic. A speech waveform is fed to a pre-amplifier PA having automatic gain control 26, the high, middle (viz. 250-3000 c/s.) and low frequencies being routed to a fricative selector FS, harmonic locator HL and envelope peak detector EPD respectively. In the ease of voiced sound, the detector EPD (which responds to the fundamental) starts a scan ring SCR and a time base ring TBR, after delays. The scan ring SCR distributes the pulses from the harmonic locator HL (presence of a particular harmonic being indicated by presence of a pulse at a corresponding time) to integrating pulse shapers (having a threshold) IPS2-IPS8, the digital outputs from which are set into rows 2-8 of a storage matrix SM, at column positions successively selected by the time base ring TBR. On the other hand, in the case of fricative sound, an integrating pulse shaper IPS 1, fed from the fricative selector FS, starts the time base ring TBR, after a delay, to set successive values of the output of IPS1 into row 1 of the storage matrix SM. The harmonic locator HL consists of a peak detector feeding a univibrator (to produce standard pulses). The delays in activation of the rings are to prevent a pulse from the harmonic locator HL representing the fundamental from being used.

    Abstract translation: 1,154,129。 语音识别。 国际商业机器公司1967年12月20日[1966年12月29日]，第57991/67号。 标题G4R。 在语音分析装置中，检测到谐波和基波，后者控制装置的检测提供表示每个谐波发生时间的信号。 语音波形被馈送到具有自动增益控制26的前置放大器PA，高中间（即250-3000c / s），低频被路由到摩擦选择器FS，谐波定位器HL和包络峰值检测器 EPD。 在声音的轻松中，检测器EPD（其响应于基准）在延迟之后启动扫描环SCR和时基环TBR。 扫描环SCR将来自谐波定位器HL（存在特定谐波的脉冲存在于相应的时间）的脉冲分配给积分脉冲整形器（具有阈值）IPS2-IPS8，其数字输出被设置 在由时基环TBR连续选择的列位置处的存储矩阵SM的行2-8中。 另一方面，在摩擦音的情况下，从摩擦选择器FS馈送的积分脉冲整形器IPS1在延迟之后启动时基环TBR，以将IPS1的输出的连续值设置为 存储矩阵SM。 谐波定位器HL由馈送单谐振器的峰值检测器（产生标准脉冲）组成。 环的激活延迟是为了防止来自表示基准的谐波定位器HL的脉冲被使用。

Patent Agency Ranking