Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
automatic speech recognition full report
Post: #1


Automatic Speech Recognition

Automatic speech recognition

¢ What is the task
¢ What are the main difficulties
¢ How is it approached
¢ How good is it
¢ How much better could it be

What is the task

¢ Getting a computer to understand spoken language
¢ By understand we might mean
“ React appropriately
“ Convert the input speech into another medium, e.g. text
¢ Several variables impinge on this (see later)

How do humans do it

¢ Articulation produces
¢ sound waves which
¢ the ear conveys to the brain
¢ for processing

How might computers do it

¢ Digitization
¢ Acoustic analysis of the speech signal
¢ Linguistic interpretation

Whatâ„¢s hard about that

¢ Digitization
“ Converting analogue signal into digital representation
¢ Signal processing
“ Separating speech from background noise
¢ Phonetics
“ Variability in human speech
¢ Phonology
“ Recognizing individual sound distinctions (similar phonemes)
¢ Lexicology and syntax
“ Disambiguating homophones
“ Features of continuous speech
¢ Syntax and pragmatics
“ Interpreting prosodic features
¢ Pragmatics
“ Filtering of performance errors (disfluencies)


¢ Analogue to digital conversion
¢ Sampling and quantizing
¢ Use filters to measure energy levels for various points on the frequency spectrum
¢ Knowing the relative importance of different frequency bands (for speech) makes this process more efficient
¢ E.g. high frequency sounds are less informative, so can be sampled using a broader bandwidth (log scale)

Separating speech from background noise

¢ Noise cancelling microphones
“ Two mics, one facing speaker, the other facing away
“ Ambient noise is roughly same for both mics
¢ Knowing which bits of the signal relate to speech
“ Spectrograph analysis

Variability in individualsâ„¢ speech

¢ Variation among speakers due to
“ Vocal range (f0, and pitch range “ see later)
“ Voice quality (growl, whisper, physiological elements such as nasality, adenoidality, etc)
“ ACCENT !!! (especially vowel systems, but also consonants, allophones, etc.)
¢ Variation within speakers due to
“ Health, emotional state
“ Ambient conditions
¢ Speech style: formal read vs spontaneous

Speaker-(in)dependent systems

¢ Speaker-dependent systems
“ Require training to teach the system your individual idiosyncracies
¢ The more the merrier, but typically nowadays 5 or 10 minutes is enough
¢ User asked to pronounce some key words which allow computer to infer details of the user™s accent and voice
¢ Fortunately, languages are generally systematic
“ More robust
“ But less convenient
“ And obviously less portable
¢ Speaker-independent systems
“ Language coverage is reduced to compensate need to be flexible in phoneme identification
“ Clever compromise is to learn on the fly

Identifying phonemes

¢ Differences between some phonemes are sometimes very small
“ May be reflected in speech signal (eg vowels have more or less distinctive f1 and f2)
“ Often show up in coarticulation effects (transition to next sound)
¢ e.g. aspiration of voiceless stops in English
“ Allophonic variation

Disambiguating homophones

¢ Mostly differences are recognised by humans by context and need to make sense
Itâ„¢s hard to wreck a nice beach
What dimeâ„¢s a neckâ„¢s drain to stop port
¢ Systems can only recognize words that are in their lexicon, so limiting the lexicon is an obvious ploy
¢ Some ASR systems include a grammar which can help disambiguation
¢ Discontinuous speech much easier to recognize
¢ Single words tend to be pronounced more clearly
¢ Continuous speech involves contextual coarticulation effects
¢ Weak forms
¢ Assimilation
¢ Contractions

Interpreting prosodic features

¢ Pitch, length and loudness are used to indicate stress
¢ All of these are relative
“ On a speaker-by-speaker basis
“ And in relation to context
¢ Pitch and length are phonemic in some languages


¢ Pitch contour can be extracted from speech signal
“ But pitch differences are relative
“ One man™s high is another (wo)man™s low
“ Pitch range is variable
¢ Pitch contributes to intonation
“ But has other functions in tone languages
¢ Intonation can convey meaning

¢ Length is easy to measure but difficult to interpret
¢ Again, length is relative
¢ It is phonemic in many languages
¢ Speech rate is not constant “ slows down at the end of a sentence

Template-based approach

¢ Hard to distinguish very similar templates
¢ And quickly degrades when input differs from templates
¢ Therefore needs techniques to mitigate this degradation:
“ More subtle matching techniques
“ Multiple templates which are aggregated
¢ Taken together, these suggested ¦

Rule-based approach

¢ Use knowledge of phonetics and linguistics to guide search process
¢ Templates are replaced by rules expressing everything (anything) that might help to decode:
“ Phonetics, phonology, phonotactics
“ Syntax
“ Pragmatics

Statistics-based approach

¢ Can be seen as extension of template-based approach, using more powerful mathematical and statistical tools
¢ Sometimes seen as anti-linguistic approach
“ Fred Jelinek (IBM, 1988): Every time I fire a linguist my system improves
¢ Collect a large corpus of transcribed speech recordings
¢ Train the computer to learn the correspondences (machine learning)
¢ At run time, apply statistical processes to search through the space of all possible solutions, and pick the statistically most likely one
Machine learning

¢ Acoustic and Lexical Models
“ Analyse training data in terms of relevant features
“ Learn from large amount of data different possibilities
¢ different phone sequences for a given word
¢ different combinations of elements of the speech signal for a given phone/phoneme
“ Combine these into a Hidden Markov Model expressing the probabilities

The Noisy Channel Model

¢ Use the acoustic model to give a set of likely phone sequences
¢ Use the lexical and language models to judge which of these are likely to result in probable word sequences
¢ The trick is having sophisticated algorithms to juggle the statistics
¢ A bit like the rule-based approach except that it is all learned automatically from data


¢ Funders have been very keen on competitive quantitative evaluation
¢ Subjective evaluations are informative, but not cost-effective
¢ For transcription tasks, word-error rate is popular (though can be misleading: all words are not equally important)
¢ For task-based dialogues, other measures of understanding are needed

Comparing ASR systems

¢ Factors include
“ Speaking mode: isolated words vs continuous speech
“ Speaking style: read vs spontaneous
“ Enrollment: speaker (in)dependent
“ Vocabulary size (small <20 ¦ large > 20,000)
“ Equipment: good quality noise-cancelling mic ¦ telephone
“ Size of training set (if appropriate) or rule set
“ Recognition method

Remaining problems

¢ Robustness “ graceful degradation, not catastrophic failure
¢ Portability “ independence of computing platform
¢ Adaptability “ to changing conditions (different mic, background noise, new speaker, new task domain, new language even)
¢ Language Modelling “ is there a role for linguistics in improving the language models
¢ Confidence Measures “ better methods to evaluate the absolute correctness of hypotheses.
¢ Out-of-Vocabulary (OOV) Words “ Systems must have some method of detecting OOV words, and dealing with them in a sensible way.
¢ Spontaneous Speech “ disfluencies (filled pauses, false starts, hesitations, ungrammatical constructions etc) remain a problem.
¢ Prosody “Stress, intonation, and rhythm convey important information for word recognition and the user's intentions (e.g., sarcasm, anger)
¢ Accent, dialect and mixed language “ non-native speech is a huge problem, especially where code-switching is commonplace
Post: #2
plz give me detailed description of this report as early as possible
Post: #3
Thank you for your entire effort on this web page. Betty takes pleasure in setting aside time for internet research and it is easy to see why. We all learn all relating to the lively medium you render both useful and interesting information via this blog and as well inspire response from website visitors about this subject matter while our daughter has always been learning a great deal. Enjoy the remaining portion of the year. Your doing a tremendous job. ellipticals for home
Post: #4
to get information about the topic"speech recognition project source code in c#" related topic refer the link bellow

Important Note..!

If you are not satisfied with above reply ,..Please


So that we will collect data for you and will made reply to the request....OR try below "QUICK REPLY" box to add a reply to this page
Popular Searches: overview of speech recognition report, full seminar report on speech recognition, speech recognition full report using vhdl code, automatic speech recognition tutorial, speech recognition seminar report in electronic, automatic speech recognition dsp, full automatic sensorborewels,

Quick Reply
Type your reply to this message here.

Image Verification
Image Verification
(case insensitive)
Please enter the text within the image on the left in to the text box below. This process is used to prevent automated posts.

Possibly Related Threads...
Thread: Author Replies: Views: Last Post
  software defined radio full report computer science technology 14 9,328 19-10-2015 02:51 PM
Last Post: seminar report asees
  synthetic aperture radar system full report computer science technology 11 10,352 25-03-2015 11:07 AM
Last Post: seminar report asees
  satrack full report computer science technology 8 13,797 21-07-2013 08:32 AM
Last Post: Guest
  wearable biosensors full report computer science technology 3 11,228 01-03-2013 11:47 AM
Last Post: seminar details
  Power Point Tracking for Photovoltaic Systems full report computer science technology 1 2,951 19-01-2013 12:51 PM
Last Post: seminar details
  face recognition technology seminar report applied electronics 1 4,726 14-01-2013 01:07 PM
Last Post: seminar details
  robotics and its applications full report computer science technology 5 12,167 21-12-2012 11:58 AM
Last Post: seminar details
  embedded configurable operating system full report project reporter 1 3,966 11-12-2012 01:32 PM
Last Post: seminar details
  adaptive missle guidance full report computer science technology 1 3,536 10-12-2012 03:28 PM
Last Post: seminar details
  Wireless Battery Charger Chip for Smart-Card Applications full report project topics 6 5,610 09-11-2012 11:53 AM
Last Post: seminar details