![Steps for EPD
26/34
•Removal of noise by subtracting the signal values with that of noise
• Word extraction
steps –
1. ITU [Upper energy threshold]
2. ITL [Lower energy threshold]
3. IZCT [Zero crossing rate threshold ]](https://image.slidesharecdn.com/automaticspeechrecognition-changed-100709112514-phpapp02/85/Automatic-speech-recognition-26-320.jpg)
Automatic speech recognition
- 2. 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? 2/34
- 3. 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 3/34
- 4. How do humans do it? Articulation produces sound waves which the ear conveys to the brain for processing 4/34
- 5. How might computers do it? Digitization Acoustic analysis of the speech signal Linguistic interpretation 5/34 Acoustic waveform Acoustic signal Speech recognition
- 7. What’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) 7/34
- 8. Digitization 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) 8/34
- 9. 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 9/34
- 10. Variability in individuals’ 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 10/34
- 11. 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 11/34
- 12. (Dis)continuous speech Discontinuous speech much easier to recognize Single words tend to be pronounced more clearly Continuous speech involves contextual coarticulation effects Weak forms Assimilation Contractions 12/34
- 13. Performance errors Performance “errors” include Non-speech sounds Hesitations False starts, repetitions Filtering implies handling at syntactic level or above Some disfluencies are deliberate and have pragmatic effect – this is not something we can handle in the near future 13/34
- 15. Template-based approach Store examples of units (words, phonemes), then find the example that most closely fits the input Extract features from speech signal, then it’s “just” a complex similarity matching problem, using solutions developed for all sorts of applications OK for discrete utterances, and a single user 15/34
- 16. 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 … 16/34
- 17. Neural Network based approach 17/34
- 18. Statistics-based approach 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 18/34
- 19. Statistics based approach 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 19/34
- 20. HMMs for some words 20/34
- 21. Identify individual phonemes Identify words Identify sentence structure and/or meaning 21/34
- 22. SPEECH RECOGNITION BLOCK DIAGRAM 22/34
- 23. BLOCK DIAGRAM DESCRIPTION 23/34 Speech Acquisition Unit •It consists of a microphone to obtain the analog speech signal •The acquisition unit also consists of an analog to digital converter Speech Recognition Unit •This unit is used to recognize the words contained in the input speech signal. •The speech recognition is implemented in MATLAB with the help of •template matching algorithm Device Control Unit •This unit consists of a microcontroller, the ATmega32, to control the various appliances •The microcontroller is connected to the PC via the PC parallel port •The microcontroller then reads the input word and controls the device connected to it accordingly.
- 25. END-POINT DETECTION 25/34 • The accurate detection of a word's start and end points means that subsequent processing of the data can be kept to a minimum by processing only the parts of the input corresponding to speech. •We will use the endpoint detection algorithm proposed by Rabiner and Sambur. This algorithm is based on two simple time-domain measurements of the signal - the energy and the zero crossing rate. The algorithm should tackle the following cases:- 1. Words which begin with or end with a low energy phoneme 2. Words which end with a nasal 3. Speakers ending words with a trailing off in intensity or short breath
- 26. Steps for EPD 26/34 •Removal of noise by subtracting the signal values with that of noise • Word extraction steps – 1. ITU [Upper energy threshold] 2. ITL [Lower energy threshold] 3. IZCT [Zero crossing rate threshold ]
- 27. Feature Extraction Input data to the algorithm is usually too large to be processed Input data is highly redundant Raw analysis requires high computational powers and large amounts of memory Thus, removing the redundancies and transforming the data into a set of features DCT based Mel Cepstrum 27/34
- 28. DCT Based MFCC • Take the Fourier transform of a signal. • Map the powers of the spectrum obtained above onto the mel scale, using triangular overlapping windows. • Take the logs of the powers at each of the mel frequencies. • Take the discrete cosine transform of the list of mel log powers, as if it were a signal. • The MFCCs are the amplitudes of the resulting spectrum. 28/34
- 29. MFCC Computation As Log Magnitude is real and symmetric IDFT reduces to DCT. The DCT produces highly un-correlated feature yt (m)(k). The Zero Order MFCC coefficient yt (0)(k) is approximately equal to the Log Energy of the frame. 29/34The number of MFCC co-effecients chosen were 13
- 30. Feature extraction by MFCC Processing 30/34
- 31. Dynamic Time Warping and Minimum Distance Paths measurement Isolated word recognition: • Task : • Want to build an isolated word recogniser • Method: 1. Record, parameterise and store vocabulary of reference words. 2. Record test word to be recognised and parameterize. 3. Measure distance between test word and each reference word. 4. Choose reference word ‘closest’ to test word. 31/34
- 32. 32 Words are parameterised on a frame-by-frame basis Choose frame length, over which speech remains reasonably stationary Overlap frames e.g. 40ms frames, 10ms frame shift We want to compare frames of test and reference words i.e. calculate distances between them 40ms 20m s
- 33. 33 • Hard: Number of frames won’t always correspond • Easy: Sum differences between corresponding frames Calculating Distances
- 34. 34 • Solution 1: Linear Time Warping Stretch shorter sound • Problem? Some sounds stretch more than others
- 35. 35 • Solution 2: Dynamic Time Warping (DTW) 5 3 9 7 3 4 7 4 Test Reference Using a dynamic alignment, make most similar frames correspond Find distances between two utterances using these corresponding frames
- 36. Dynamic Programming 36 Waveforms showing the utterance of the word “Open” at two different instants. The signals are not time aligned.
- 37. 37 3 5 1 x 4 x 1 x 7 4 3 x 0 x 3 x 9 3 5 x 2 x 5 x 3 2 1 x 4 x 1 x 5 1 1 x 2 x 1 x 1 2 3 4 7 4 Reference T e s t Place distance between frame r of Test and frame c of Reference in cell(r,c) of distance matrix DTW Process
- 38. Constraints Global Endpoint detection Path should be close to diagonal Local Must always travel upwards or eastwards No jumps Slope weighting Consecutive moves upwards/eastwards 38
- 39. Empirical Results : Known Speaker 39 SONY SUVARNA GEMINI HBO CNN NDTV IMAGINE ZEE CINEMA SONY 9 0 1 0 0 0 0 0 SUVARNA 0 10 0 0 0 0 0 0 GEMINI 0 0 8 0 0 0 2 0 HBO 0 0 0 10 0 0 0 0 CNN 0 0 0 0 8 0 2 0 NDTV 0 0 0 0 0 10 0 0 IMAGINE 0 0 0 0 0 0 10 0 ZEE CINEMA 0 0 0 0 0 0 1 9
- 40. Empirical Results : Unknown Speaker 40 SONY SUVARNA GEMINI HBO CNN NDTV IMAGINE ZEE CINEMA SONY 8 0 1 0 0 0 1 0 SUVARNA 0 8 0 0 0 0 0 2 GEMINI 1 0 8 0 0 0 1 0 HBO 0 0 0 10 0 0 0 0 CNN 1 0 0 0 8 0 2 0 NDTV 0 0 0 0 0 10 0 0 IMAGINE 0 0 0 0 0 0 10 0 ZEE CINEMA 0 2 0 0 0 0 0 8
- 41. Applications Medical Transcription Military Telephony and other domains Serving the disabled Further Applications • Home automation • Automobile audio systems • Telematics 41
- 43. 43/34 all speakers of the language including foreign application independent or adaptive all styles including human-human (unaware) wherever speech occurs 2015 vehicle noise radio cell phones regional accents native speakers competent foreign speakers some application– specific data and one engineer year natural human- machine dialog (user can adapt) 1995 expert years to create app– specific language model speaker independent and adaptive normal office various microphones telephone planned speech 1985 NOISE ENVIRONMENT SPEECH STYLE USER POPULATION COMPLEXITY 1975 quiet room fixed high – quality mic careful reading speaker-dep. application– specific speech and language Evolution of ASR
- 44. 44 Conclusive remarks Recorded Speech Noise Padded Gain Adjustment DC offset elimination Spectral Subtraction End Point Detection