{"resolved":true,"match_type":"uuid","source_type":"original_book","book_id":"bbcce9bd-b637-4fa6-8a80-dadabab80104","title":"数字化生存 20周年纪念版","author":null,"edition":null,"publisher":null,"year":null,"isbn":"9787121307362","requested_by":"printed_page","book":{"source_type":"original_book","book_id":"bbcce9bd-b637-4fa6-8a80-dadabab80104","id":"bbcce9bd-b637-4fa6-8a80-dadabab80104","uuid":"bbcce9bd-b637-4fa6-8a80-dadabab80104","title":"数字化生存 20周年纪念版","normalized_title":"数字化生存20周年纪念版","author":null,"edition":null,"publisher":null,"year":null,"isbn":"9787121307362","isbns":["9787121307362"],"parsed":true,"ai_readable":true,"document_type":"book","page_count":362,"navigation_url":"https://workshop.closeclaw.top/ai/book/bbcce9bd-b637-4fa6-8a80-dadabab80104","title_url":"https://workshop.closeclaw.top/ai/title/%E6%95%B0%E5%AD%97%E5%8C%96%E7%94%9F%E5%AD%98%2020%E5%91%A8%E5%B9%B4%E7%BA%AA%E5%BF%B5%E7%89%88","fixed_page_url_template":"https://workshop.closeclaw.top/ai/book/bbcce9bd-b637-4fa6-8a80-dadabab80104/page/{printed_page}","path_api_url_template":"https://workshop.closeclaw.top/api/ai/book/bbcce9bd-b637-4fa6-8a80-dadabab80104/page/{printed_page}","read_url":"https://workshop.closeclaw.top/api/ai/read/bbcce9bd-b637-4fa6-8a80-dadabab80104","page_url":"https://workshop.closeclaw.top/api/ai/books/bbcce9bd-b637-4fa6-8a80-dadabab80104/page","search_url":"https://workshop.closeclaw.top/api/ai/books/bbcce9bd-b637-4fa6-8a80-dadabab80104/search"},"page":{"printed_page":137,"pdf_page":208,"physical_index":208,"physical_index_base":0,"printed_page_status":"recognized","page_sha256":"f1407f5a0368dbd53342671c6ecfb298d2fb5b969b6840964f44e0f83864d5f5"},"chapter":"20周年纪念版 数字化生存 一本影响千百万人的互联网启蒙经典著作","content":"我们可以把语音识别的这三个方面想象成三维立体轴。在词汇轴上，需要识别的字越少，对计算机而言就越容易。假如系统事先就知道说话的人是谁，问题就更简单了。如果说话的人能一个字一个字分开发音，计算机就听得更明白，识别起来也就更容易。在这三条轴的起始点，我们可以找到少得不能再少、完全依赖于说话者语音的词汇，念出这些词汇的时候，字与字之间必须有明——显——的——停——顿。\n\n当我们顺着任何一条轴移动的时候，也就是说，增加计算机所能识别的词汇，让系统能够服务于任何说话者，或是容许字与字相连的程度越来越高，在这种情况下，每前进一步.都会使问题变得越来越困难。当到达三条轴的远端时，我们期望计算机能够识别任何人说的任何字，以及“印（任）何程度”的含混字。人们通常都认为，我们必须在两条或三条轴上都达到最远端时，语音识别系统才能对人类有用。这完全不对。\n\n让我们一个一个来考虑。谈到词汇多寡的时候，我们可能会问：多少才算多呢？500个字、5000个字还是50000个字?但这个问题其实应该是：在任何时候，计算机存储器中究竟需要多少它可以识别的单字？这个问题提示我们把词汇根据上下文分成组，这样在需要的时候就可以把大群词组放进存储器中。当我要求我的计算机接听电话时，它会把信息输入我的电子电话本。当我计划到什么地方旅行时，它则把地名输入到上面。\n\nPart 2|个性化界面\n\n137","previous_page":136,"next_page":138}