THEORY EXAMINATION (SEM–VIII) 2016-17 SPEECH PROCESSING

SECTION A – Basic Concepts of Speech Processing

Section A includes short questions that test fundamental knowledge of speech signals, signal processing, and acoustic properties of speech. 

Question (a): What is Pitch?

Answer:
Pitch is the perceptual property of sound that allows humans to classify sounds as high or low. In speech processing, pitch corresponds to the fundamental frequency of the speech signal produced by vibration of the vocal cords.

When vocal cords vibrate rapidly, the pitch becomes high. When they vibrate slowly, the pitch becomes low.

Pitch is important in speech processing because it helps identify speaker characteristics and plays a role in speech recognition and speech synthesis systems.

Question (b): Explain Acoustic Phonetics

Answer:
Acoustic phonetics is the study of the physical properties of speech sounds. It focuses on how speech signals are produced, transmitted, and received.

It analyzes properties such as:

Frequency

Amplitude

Duration

Spectral characteristics

Acoustic phonetics helps researchers understand how speech signals behave and how they can be processed digitally for applications like speech recognition and voice synthesis.

Question (c): Why is Sampling Required?

Answer:
Sampling is required to convert a continuous-time speech signal into a discrete-time signal so that it can be processed digitally by computers.

Speech signals are naturally analog, meaning they vary continuously over time. However, digital systems require discrete values. Sampling captures the amplitude of the signal at specific intervals.

According to the Nyquist sampling theorem, the sampling frequency must be at least twice the maximum frequency present in the signal to accurately reconstruct the original signal.

Question (d): Define Channel Vocoder

Answer:
A channel vocoder is a type of speech analysis and synthesis system used to compress speech signals.

It works by splitting the speech signal into multiple frequency bands using filters. Each band is analyzed to extract important features such as amplitude and pitch.

These features are transmitted instead of the entire speech waveform, which reduces the amount of data required for transmission.

Channel vocoders are widely used in speech compression and communication systems.

Question (e): What is Frequency Domain?

Answer:
The frequency domain represents a signal in terms of its frequency components rather than time.

In speech processing, signals are often analyzed in the frequency domain because it helps identify important features such as pitch, harmonics, and formants.

Techniques such as the Fourier Transform are used to convert time-domain signals into frequency-domain representations.

Question (f): Define Correlation Function with Example

Answer:
The correlation function measures the similarity between two signals or between different parts of the same signal.

In speech processing, correlation functions are used to detect repeating patterns in speech signals, such as pitch.

For example, if a speech signal repeats periodically, the correlation function will show peaks at time intervals corresponding to the pitch period.

Question (g): What is a Filter?

Answer:
A filter is a signal processing device or algorithm used to remove unwanted components from a signal.

Filters are widely used in speech processing to remove noise or isolate specific frequency bands.

Common types of filters include:

Low-pass filters

High-pass filters

Band-pass filters

Band-stop filters

Filters help improve speech quality and clarity.

Question (h): Difference Between Speech and Silence

Answer:

Speech processing systems must detect silence periods to reduce computational load and improve efficiency.

Question (i): Define Convolution with Example

Answer:
Convolution is a mathematical operation used to combine two signals to produce a third signal.

In speech processing, convolution is used to model how speech signals pass through systems such as filters or vocal tract models.

For example, if a speech signal passes through a filter, the output signal is the convolution of the input signal and the filter response.

Question (j): What is Linear Predictive Coding (LPC)?

Answer:
Linear Predictive Coding is a technique used in speech processing to represent the spectral envelope of speech signals efficiently.

LPC works by predicting the current speech sample based on past samples.

It reduces the amount of data required to represent speech while maintaining intelligibility.

LPC is widely used in:

Speech synthesis

Speech compression

Voice communication systems

SECTION B – Intermediate Concepts of Speech Processing

Section B questions focus on speech signal analysis, modeling, and parameter extraction techniques. 

Question: Sampling and Quantization in Speech Signals

Answer:
Sampling and quantization are two essential processes used to convert analog speech signals into digital form.

Sampling involves measuring the amplitude of a speech signal at regular time intervals. This converts the continuous signal into a discrete-time signal.

Quantization is the process of converting sampled amplitudes into discrete levels so they can be represented digitally.

For example, when recording speech using a microphone, the analog signal is sampled and quantized before being stored as digital audio.

Question: Digital Models for Speech Signals

Answer:
Digital models attempt to represent speech signals mathematically.

One widely used model is the source-filter model, which assumes that speech is produced by a sound source (vocal cords) and shaped by the vocal tract.

This model helps in understanding speech production and is used in speech synthesis systems.

Question: Applications of Speech Processing

Speech processing has many applications in modern technology, including:

Speech recognition systems

Voice assistants

Automatic transcription

Speaker identification

Hearing aids

Voice-controlled devices

These applications improve human-computer interaction and accessibility.

Question: Short-Term Pitch Detection

Short-term pitch detection estimates the pitch of speech signals within short time frames.

The process typically involves:

Segmenting the speech signal into frames

Computing correlation functions

Detecting peaks corresponding to pitch periods

Pitch detection is used in speech synthesis and speaker recognition.

SECTION C – Advanced Concepts of Speech Processing

Section C includes deeper theoretical concepts such as speech synthesis and Fourier analysis. 

Question: Speech Synthesis

Speech synthesis is the process of generating artificial speech using computers.

It involves converting text or symbolic information into speech signals.

Speech synthesis systems typically include:

Text analysis

Phoneme generation

Prosody generation

Speech waveform generation

Linear Predictive Coding plays an important role in speech synthesis because it efficiently models the vocal tract and generates realistic speech signals.

Question: Short-Time Fourier Analysis

Short-Time Fourier Transform (STFT) is used to analyze how the frequency components of speech signals change over time.

Because speech signals are non-stationary, analyzing them using short time windows provides better understanding of their dynamic properties.

STFT divides the signal into small frames and computes the Fourier transform for each frame.

This method helps visualize speech signals using spectrograms, which display frequency variation over time.

Question: Autocorrelation, NMSE, and Formant Estimation

Autocorrelation Method

Autocorrelation measures similarity between a signal and delayed versions of itself. It is commonly used for pitch detection.

Normalized Mean Square Error (NMSE)

NMSE measures the difference between predicted and actual speech signals. It is used to evaluate the accuracy of speech models.

Formant Estimation

Formants are resonance frequencies of the vocal tract. Estimating formants helps identify vowel sounds and is important in speech recognition systems.

Conclusion

Speech processing combines signal processing techniques with linguistic knowledge to analyze, synthesize, and recognize speech signals. Concepts such as pitch detection, sampling, filtering, and linear predictive coding play a crucial role in building speech-based technologies.

These technologies power modern applications such as voice assistants, automated transcription systems, and speech-enabled communication devices.