Spontaneous speech recognition
Spontaneous speech recognition
批准号:
15500098
负责人:
KOHDA Masaki
金额:
$2.05万
依托单位:
依托单位国家:
日本
项目类别:
Grant-in-Aid for Scientific Research (C)
财政年份:
2003
资助国家:
日本
项目状态:
已结题
起止时间:
2003 至 2005
中文摘要
我们对学术演讲任务中的自发语音识别进行了研究,得到了以下结果。(1)基于语音变异建模和无监督自适应的演讲语音识别我们关注的是自发语音中观察到的语音变异。为了引入语音变体的语境依赖性,提出了一种基于语音变体形态分析数据的语言建模新方法。在自发日语语料库(CSJ)上对该方法进行了测试,结果表明该方法的单词错误率(WER)绝对降低了4.74%。此外,引入了声学和语言模型的无监督自适应,进一步提高了识别性能。结果表明:无监督适应时,WER由无监督适应时的19.96%下降到无监督适应时的15.41%;(2)基于离散混合HMM的演讲语音识别我们研究了基于离散混合HMM (DMHMM)的噪声语音识别,发现DMHMM在环境噪声或脉冲噪声条件下的性能优于连续混合HMM。然而,尚不清楚这种方法在清洁条件下是否有效。本研究的目的是评估DMHMM系统在清洁条件下的性能。在评估中,我们决定使用“自发日语语料库”(CSJ),因为我们想将我们的系统与其他具有常见语音语料库的识别系统的性能进行比较,并澄清在这样一个更困难的任务中的性能。在识别实验中,使用3000状态dmhmm(每种状态16种混合成分)作为声学模型。用CSJ 2668个讲座的686万个单词训练了代表发音变化的语言模型,并用于识别。结果表明,该系统对10个男性主讲人的学术讲座获得了20.30%的WER,证明了该方法的有效性。少
英文摘要
We investigated spontaneous speech recognition on academic lecture task and obtained the following results.(1) Lecture speech recognition using pronunciation variant modeling and unsupervised adaptationWe focus on the pronunciation variations observed in spontaneous speech. Aiming to introduce the context-dependence of pronunciation variants, we propose a new method of language modeling based on morphological analysis data designed for pronunciation variant. The proposed method was evaluated on the Corpus of Spontaneous Japanese (CSJ) and achieved the decrease in word error rate (WER) by 4.74% absolute. In addition, unsupervised adaptation of both acoustic and language models was introduced to improve the recognition performance further. The results showed the decrease in WER from 19.96% without adaptation to 15.41% with unsupervised adaptation.(2) Lecture speech recognition using discrete-mixture HMMsWe have investigated noisy speech recognition by using discrete-mixture HMM (DMHMM), … More and found that the performance of DMHMM overcame that of continuous-mixture HMM under environmental noise conditions or impulsive noise conditions. However, it is not clear whether this method is effective in clean conditions. The aim of this investigation is to evaluate the performance of the DMHMM system in clean conditions. In evaluation, we decided to use the "Corpus of Spontaneous Japanese" (CSJ) because we want to compare the performance of our system with that of other recognition systems with common speech corpus, and clarify the performance in such a more difficult task. In the recognition experiments, 3000-state DMHMMs (16 mixture components per state) were used as acoustic models. The language model which represents the pronunciation variety was trained by using 6.86 million words from 2668 lectures in CSJ and was used for recognition. As a result, the system obtained 20.30% WER for 10 academic lectures uttered by male speakers and demonstrated the effectiveness of the proposed method. Less
期刊论文(75)
专著(0)
科研奖励(0)
会议论文
登录
查看更多内容
話者ベクトルを用いた雑音下話者認識手法の検討
基于说话人向量的噪声下说话人识别方法研究
DOI:
--
发表时间:
2006
期刊:
情報処理学会東北支部研究会 05-5-A1-2
影响因子:
--
作者:
[R.Tsutsumi, M.Katoh, T.Kosaka, M.Kohda, 加藤正治, 阿部拓也, 遠藤大悟, 熊倉拓哉, 赤津達也]
通讯作者:
赤津達也
Rebust Speech Recognition Using Discrete-Mixture HMMs
使用离散混合 HMM 重构语音识别
DOI:
--
发表时间:
2005
期刊:
IEICE Trans. on Information and Systems(電子情報通信学会英文論文誌) E88-D,12
影响因子:
--
作者:
[T.Kosaka, M.Katoh, M.Kohda, 小坂哲夫, 阿部拓也, 小坂哲夫, 阿部拓也, 小坂哲夫]
通讯作者:
小坂哲夫
松本 和樹: "分散音声認識のクライアントにおけるマイク特性変動の除去"情報処理学会 東北支部研究会. 03-5-B2-2. 1-8 (2004)
Kazuki Matsumoto:“分布式语音识别客户端中麦克风特性波动的消除”日本信息处理学会东北分会研究组 03-5-B2-2 (2004)。
DOI:
--
发表时间:
期刊:
影响因子:
--
作者:
[]
通讯作者:
小坂 哲夫: "Noisy speech recognition with discrete-mixture HMMs based on MAP estimation"18th International Congress on Acoustics. Tu. P2.8. (2004)
Tetsuo Kosaka:“基于 MAP 估计的离散混合 HMM 的噪声语音识别”第 18 届国际声学大会 (2004)。
DOI:
--
发表时间:
期刊:
影响因子:
--
作者:
[]
通讯作者:
発音変形依存モデルを用いた講演音声認識
使用发音转换依赖模型的讲座语音识别
DOI:
--
发表时间:
2006
期刊:
電子情報通信学会論文誌 J89-D,2
影响因子:
--
作者:
[堤怜介]
通讯作者:
堤怜介
共 37 条
Large-vocabulary continuous speech recognition on spontaneous speech task
-
批准号:18500126
-
项目类别:Grant-in-Aid for Scientific Research (C)
-
资助金额:$1.22万
-
财政年份:2006
-
负责人:KOHDA Masaki
-
依托单位:
Large Vocabulary Continuous Speech Recognition System on Japanese Newspaper Reading Task
-
批准号:10680368
-
项目类别:Grant-in-Aid for Scientific Research (C)
-
资助金额:$2.11万
-
财政年份:1998
-
负责人:KOHDA Masaki
-
依托单位:
Algorithm of Spontaneous Speech Recognition Based on A^<**> Search
-
批准号:07680379
-
项目类别:Grant-in-Aid for Scientific Research (C)
-
资助金额:$1.09万
-
财政年份:1995
-
负责人:KOHDA Masaki
-
依托单位:
Speech Recognition Based on Intelligent Beam Search Algorithm
-
批准号:01460254
-
项目类别:Grant-in-Aid for General Scientific Research (B)
-
资助金额:$4.42万
-
财政年份:1989
-
负责人:KOHDA Masaki
-
依托单位:
海外基金