语音的基本概念--译自CMU sphinx

语音的基本概念--译自CMU sphinx





       这是CMU sphinx语音识别系统wiki的第一部分,主要是介绍语音的一些基本概念的。我试着翻译了一下。英语水平受限,翻译难免出错,请各位不吝指点!呵呵


Basic concepts of speech


Speech is a complex phenomenon. People rarely understand how is it produced and perceived. The naive perception is often that speech is built with words, and each word consists of phones. The reality is unfortunately very different. Speech is a dynamic process without clearly distinguished parts. It's always useful to get a sound editor and look into the recording of the speech and listen to it. Here is for example the speech recording in an audio editor.



All modern descriptions of speech are to some degree probabilistic. That means that there are no certain boundaries between units, or between words. Speech to text translation and other applications of speech are never 100% correct. That idea is rather unusual for software developers, who usually work with deterministic systems. And it creates a lot of issues specific only to speech technology.



Structure of speech


In current practice, speech structure is understood as follows:


Speech is a continuous audio stream where rather stable states mix with dynamically changed states. In this sequence of states, one can define more or less similar classes of sounds, or phones. Words are understood to be built of phones, but this is certainly not true. The acoustic properties of a waveform corresponding to a phone can vary greatly depending on many factors - phone context, speaker, style of speech and so on. The so called coarticulation协同发音 makes phones sound very different from their canonical” representation. Next, since transitions between words are more informative than stable regions, developers often talk about diphones - parts of phones between two consecutive phones. Sometimes developers talk about subphonetic units - different substates of a phone. Often three or more regions of a different nature can easily be found.

The number three is easily explained. The first part of the phone depends on its preceding phone, the middle part is stable, and the next part depends on the subsequent phone. That's why there are often three states in a phone selected for HMM recognition.



协同发音(指的是一个音受前后相邻音的影响而发生变化,从发声机理上看就是人的发声器官在一个音转向另一个音时其特性只能渐变,从而使得后一个音的频谱与其他条件下的频谱产生差异。)的存在使得音素的感知与标准不一样,所以我们需要根据上下文来辨别音素。将一个音素划分为几个亚音素单元。如:数字“three”,音素的第一部分与在它之前的音素存在关联,中间部分是稳定的部分,而最后一部分则与下一个音素存在关联,这就是为什么在用HMM模型做语音识别时,选择音素的三状态HMM模型。(上下文相关建模方法在建模时考虑了这一影响,从而使模型能更准确地描述语音,只考虑前一音的影响的称为Bi-Phone,考虑前一音和后一音的影响的称为 Tri-Phone。)

Sometimes phones are considered in context. There are triphones or even quinphones. But note that unlike phones and diphones, they are matched with the same range in waveform as just phones. They just differ by name. That's why we prefer to call this object senone. A senone's dependence on context could be more complex than just left and right context. It can be a rather complex function defined by a decision tree, or in some other way.


Next, phones build subword units, like syllables. Sometimes, syllables are defined as “reduction-stable entities”. To illustrate, when speech becomes fast, phones often change, but syllables remain the same. Also, syllables are related to intonational contour. There are other ways to build subwords - morphologically-based in morphology-rich languages or phonetically-based. Subwords are often used in open vocabulary speech recognition.


Subwords form words. Words are important in speech recognition because they restrict combinations of phones significantly. If there are 40 phones and an average word has 7 phones, there must be 40^7 words. Luckily, even a very educated person rarely uses more then 20k words in his practice, which makes recognition way more feasible.


Words and other non-linguistic sounds, which we call fillers (breath, um, uh, cough), form utterances. They are separate chunks of audio between pauses. They don't necessary match sentences, which are more semantic concepts.


On the top of this, there are dialog acts like turns, but they go beyond the purpose of the document.


Recognition process


The common way to recognize speech is the following: we take waveform, split it on utterances by silences then try to recognize what's being said in each utterance. To do that we want to take all possible combinations of words and try to match them with the audio. We choose the best matching combination. There are few important things in this match.




First of all it's a concept of features. Since number of parameters is large, we are trying to optimize it. Numbers that are calculated from speech usually by dividing speech on frames. Then for each frame of length typically 10 milliseconds we extract 39 numbers that represent the speech. That's called feature vector. The way to generates numbers is a subject of active investigation, but in simple case it's a derivative from spectrum.



Second it's a concept of the model. Model describes some mathematical object that gathers common attributes of the spoken word. In practice, for audio model of senone is gaussian mixture of it's three states - to put it simple, it's a most probable feature vector. From concept of the model the following issues raised - how good does model fits practice, can model be made better of it's internal model problems, how adaptive model is to the changed conditions.



Third, it's a matching process itself. Since it would take a huge time more than universe existed to compare all feature vectors with all models, the search is often optimized by many tricks. At any points we maintain best matching variants and extend them as time goes producing best matching variants for the next frame.





According to the speech structure, three models are used in speech recognition to do the match:

An acoustic model contains acoustic properties for each senone. There are context-independent models that contain properties (most probable feature vectors for each phone) and context-dependent ones (built from senones with context).

声学模型acoustic model


A phonetic dictionary contains a mapping from words to phones. This mapping is not very effective. For example, only two to three pronunciation variants are noted in it, but it's practical enough most of the time. The dictionary is not the only variant of mapper from words to phones. It could be done with some complex function learned with a machine learning algorithm.

语音学字典phonetic dictionary



A language model is used to restrict word search. It defines which word could follow previously recognized words (remember that matching is a sequential process) and helps to significantly restrict the matching process by stripping words that are not probable. Most common language models used are n-gram language models-these contain statistics of word sequences-and finite state language models-these define speech sequences by finite state automation, sometimes with weights. To reach a good accuracy rate, your language model must be very successful in search space restriction. This means it should be very good at predicting the next word. A language model usually restricts the vocabulary considered to the words it contains. That's an issue for name recognition. To deal with this, a language model can contain smaller chunks like subwords or even phones. Please note that search space restriction in this case is usually worse and corresponding recognition accuracies are lower than with a word-based language model.

语言模型 language model


Those three entities are combined together in an engine to recognize speech. If you are going to apply your engine for some other language, you need to get such structures in place. For many languages there are acoustic models, phonetic dictionaries and even large vocabulary language models available for download.



Other concepts used


A Lattice is a directed graph that represents variants of the recognition. Often, getting the best match is not practical; in that case, lattices are good intermediate formats to represent the recognition result.


N-best lists of variants are like lattices, though their representations are not as dense as the lattice ones.

N-best listslattices有点像,但是它没有lattices那么密集(也就是保留的结果没有lattices多)。(N-best搜索和多遍搜索:为在搜索中利用各种知识源,通常要进行多遍搜索,第一遍使用代价低的知识源(如声学模型、语言模型和音标词典),产生一个候选列表或词候选网格,在此基础上进行使用代价高的知识源(如4阶或5阶的N-Gram4阶或更高的上下文相关模型)的第二遍搜索得到最佳路径。)

Word confusion networks (sausages) are lattices where the strict order of nodes is taken from lattice edges.


Speech database - a set of typical recordings from the task database. If we develop dialog system it might be dialogs recorded from users. For dictation system it might be reading recordings. Speech databases are used to train, tune and test the decoding systems.


Text databases - sample texts collected for language model training and so on. Usually, databases of texts are collected in sample text form. The issue with collection is to put present documents (PDFs, web pages, scans) into spoken text form. That is, you need to remove tags and headings, to expand numbers to their spoken form, and to expand abbreviations.

文本数据库-为了训练语言模型而收集的文本。一般是以样本文本的方式来收集形成的。而收集过程存在一个问题就是误把PDFs, web pages, scans等现成文档也当成口语文本的形式放进数据库中。所以,我们就需要把这些文件带进数据库里面的标签和文件头去掉,还有把数字展开为它们的语音形式(例如1展开为英文的one或者汉语的yi),另外还需要把缩写给扩大还原为完整单词。


What is optimized


When speech recognition is being developed, the most complex issue is to make search precise (consider as many variants to match as possible) and to make it fast enough to not run for ages. There are also issues with making the model match the speech since models aren't perfect.


Usually the system is tested on a test database that is meant to represent the target task correctly.


The following characteristics are used:


Word error rate. Let we have original text and recognition text of length of N words. From them the I words were inserted D words were deleted and S words were substituted Word error rate is

WER = (I + D + S) / N

WER is usually measured in percent.

单词错误率:我们有一个N个单词长度的原始文本和识别出来的文本。(对单词串进行识别难免有词的插入,替换和删除的误识)I代表被插入的单词个数,D代表被删除的单词个数,S代表被替换的单词个数,那么单词错误率就定义为:WER = (I + D + S) / N


Accuracy. It is almost the same thing as word error rate, but it doesn't count insertions.

Accuracy = (N - D - S) / N

Accuracy is actually a worse measure for most tasks, since insertions are also important in final results. But for some tasks, accuracy is a reasonable measure of the decoder performance.

准确度。它和单词错误率大部分是相似的,但是它不计算插入单词的个数,它定义为:Accuracy = (N - D - S) / N


Speed. Suppose the audio file was 2 hours and the decoding took 6 hours. Then speed is counted as 3xRT.


ROC curves. When we talk about detection tasks, there are false alarms and hits/misses; ROC curves are used. A curve is a graphic that describes the number of false alarms vs number of hits, and tries to find optimal point where the number of false alarms is small and number of hits matches 100%.


There are other properties that aren't often taken into account, but still important for many practical applications. Your first task should be to build such a measure and systematically apply it during the system development. Your second task is to collect the test database and test how does your application perform.


  • 30
  • 83
    觉得还不错? 一键收藏
  • 8


  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
评论 8




当前余额3.43前往充值 >
领取后你会自动成为博主和红包主的粉丝 规则
钱包余额 0