Exploring Voice Signal Processing

#

Introduction

#

In AI projects that involve human voice, such as mood detection, speaker identification, language recognition, and speech-to-text transcription, it’s essential to understand the characteristics of vocal frequencies and how sound behaves. This knowledge enables data scientists and engineers to develop effective models for tasks like real-time translation, sentiment analysis, and biometric recognition.

Understanding Human Hearing and Animal Comparisons

#

The average human ear can detect sound frequencies ranging from approximately 20 Hz to 20,000 Hz (20 kHz). However, sensitivity varies among individuals, and many people, especially as they age, may have difficulty hearing higher frequencies . Sounds below 20 Hz are termed infrasound and are generally inaudible to humans, though they can sometimes be felt as vibrations .

Dogs, on the other hand, have a broader hearing range, approximately from 40 Hz to 60,000 Hz, allowing them to perceive higher-pitched sounds that are beyond human capabilities .

The Nature of Sound Waves

#

Sound is a mechanical wave, meaning it requires a medium (such as air, water, or solids) to propagate. It cannot travel through a vacuum . Key properties of sound waves include:

• Frequency (Hz): Number of oscillations per second, determining the pitch of the sound.
• Wavelength (m): Distance between successive crests or troughs of the wave.
• Speed (m/s): Rate at which the wave travels through a medium; varies depending on the medium’s properties.
• Amplitude: Height of the wave, related to the loudness of the sound.
• Phase: The position of a point in time on a waveform cycle.
• Energy: Proportional to the square of the amplitude; higher amplitude means more energy .

Exploring Key Questions in Voice Signal Processing

#

To delve deeper into voice signal processing, consider the following questions:

• How can computers generate realistic sounds without physical instruments?

  Computers use digital signal processing (DSP) techniques to synthesize sounds by manipulating waveforms and frequencies, enabling the creation of virtual instruments and voices.

• Can computers produce entirely new sounds not found in nature?

  Yes, through sound synthesis methods like additive, subtractive, and granular synthesis, computers can generate unique sounds that don’t occur naturally.

• What mechanisms in the human vocal cords allow for distinct syllables and words?

  The human vocal apparatus, including the vocal cords, tongue, and lips, shapes airflow and vibrations to produce a wide range of sounds and articulate speech.

• How does the human ear distinguish between different speakers or identify gender and age?

  Variations in pitch, timbre, and speech patterns allow listeners to differentiate between speakers and infer characteristics like gender and age.

• How can we detect a person’s mood from their voice?

  Emotional states influence speech patterns, including tone, pitch, and rhythm, which can be analyzed to assess mood and sentiment.

Understanding these aspects is crucial for developing AI systems capable of interpreting and processing human speech effectively.

Exploring Musical Notes

#

What is octave?

#

The word “octave” comes from the Latin “octavus”, meaning eighth.

In a major scale, there are 8 notes:

C – D – E – F – G – A – B – C
(That’s 8 notes: C is counted twice—start and end)

So the 8th note is called the octave — it’s the same pitch class, but doubled in frequency.

Example on Piano:

If you play Middle C (C4) and then the next higher C (C5), that’s 1 octave apart.

There are 12 keys between them (7 white + 5 black = 12 semitones). |

Finding your voice type

#

If you want to find your voice you an android app like Riyaz. Sing ah…. at the deepest level and see what note is identified by the app. Then you can sing ee… and highest note and see what note is identified by the app. The difference between these is your vocal range. Having volcal range like 4, 4.5, 5, 5.5, 6 is extremely exceptional. Keep in mind holding the breath and maintaining that sound of ah or ee for long period of time without crack is must.

Example:

• Your lowest note (e.g., F2)
• Your highest note (e.g., C#7)

Then count the number of semitones between those two notes (or just use the vocal range calculator of Riyaz app).

Determine Your Voice Type (optional, based on gender)

If you start around F2 and go up 4.5 octaves, you may span Bass to Soprano territory — rare and impressive! It means you can sing all kind of music.

Tips to Register Notes: You can use these tips to improve your singing.

1. Chest Voice (Lower range):

• Practice descending scales on vowels like “ah” or “uh”
• Use lip trills to loosen tension
• Stay relaxed and don’t push too hard for deep notes

2. Mixed Voice (Middle):

• Try 5-tone scales blending chest and head voice
• Use consonants like “gee” or “nay” to help with balance
• Use a straw in water for SOVT (semi-occluded vocal tract) warm-ups

3. Head Voice & Falsetto (Higher):

• Start with siren exercises (“woo” or “ooh” slides)
• Focus on ease and control rather than volume
• Avoid pushing chest voice into the high range — mix instead

4. Whistle Register (if you go above C6/C7):

• Not necessary for most singers, but possible with safe, slow practice
• Begin with squeaky, soft “eee” sounds
• Requires careful technique and is best learned with a coach

Connection between Indian and Western Musical notes.

#

The Western notes A, B, C, D, E, F, G do correspond to the Sa Re Ga Ma Pa Dha Ni (Sargam) notes in Indian classical music, but the mapping depends on which note is chosen as the tonic (Sa).

Key Concept:

#

In Indian music, Sa is movable — like the tonic (root note) in a scale — and all other notes are relative to it.

In Western music, C is usually treated as the default starting point, especially in major scales (C Major = no sharps or flats).

Common Mapping (if C = Sa):

#

If D = Sa, then:

#

• Re = E
• Ga = F♯
• and so on…

So, the Western note that corresponds to “Sa” depends on your chosen pitch.

Indian music also has Komal (flat) and Tivra (sharp) versions:

• Re, Ga, Dha, Ni can be Komal (flat)
• Ma can be Tivra (sharp)

If your natural voice note is F2 then what does that mean?

#

It means musical note (F2) that corresponds to the fundamental frequency of your speaking voice.

• F2 is a note in the second octave on the musical scale.
• It has a frequency of approximately 87.31 Hz.
• It’s quite low, typically found in the bass range for male voices.

What value does it carry for singing?

#

1. Voice Type Indicator:

   • A natural speaking pitch of F2 suggests you likely have a low-pitched voice.
   • This is common in bass or bass-baritone voice types.
   • It means your comfort zone in singing (tessitura) is likely lower, and your range may extend well into the low notes.

2. Range Potential:

   • Singers often have a usable singing range that spans about 1.5 to 3 octaves above their speaking pitch.
   • If F2 is your speaking pitch, you might comfortably sing from E2 or F2 up to F4 or even higher, depending on your training.

3. Timbre and Style:

   • A lower natural pitch gives a richer, deeper timbre, which is great for:

     • Classical bass/baritone roles
     • Jazz
     • Blues
     • Some pop genres

   • However, singing high tenor or soprano parts would likely be challenging without extensive training.

In summary, Your natural speaking pitch (F2) gives a rough idea of your vocal range, tessitura, and voice type. While it doesn’t define your singing quality directly, it helps vocal coaches tailor training to maximize your natural strengths.

Connection between note, frequcy and Octave Connection

#

Here’s a table showing the frequencies (in Hz) of all 12 notes in octaves 1 to 4. These are based on A4 = 440 Hz, the standard tuning pitch used in Western music.

🎵 Notes & Frequencies by Octave (A4 = 440 Hz)

#

Each note is 12th-SQRT(2) of previous note. 12th-SQRT(2) = 1.0594 (approx) C1 = 32.79, D = 1.0594 x 32.79 = 34.72

🧠 Key Observations:

#

• Each octave doubles the frequency of the previous.
• For example:
  C2 (65.41 Hz) × 2 = C3 (130.81 Hz)
  A3 (220 Hz) × 2 = A4 (440 Hz)

What is natural frequency of human body?

#

The natural or resonant frequency of the human body refers to the specific frequencies at which various parts of the body tend to vibrate when subjected to external forces. Understanding these frequencies is crucial in fields like biomechanics, occupational health, and audio engineering.

When considered as a whole, has a fundamental resonant frequency typically ranging between 5 and 10 Hz. This low-frequency range is significant because exposure to vibrations at these frequencies can lead to discomfort or health issues, especially in occupational settings involving machinery or vehicles. For instance, studies have shown that whole-body vibrations around 10 Hz can cause resonance effects that may impact comfort and performance.

Beyond mechanical resonance, some alternative medicine practices suggest that human cells and tissues have vibrational frequencies in the megahertz (MHz) range. For example, it’s claimed that a healthy human body resonates between 62–78 MHz, and that frequencies below this range may be associated with illness.

Different parts of the human body have their own natural frequencies:

• Head: Approximately 20–30 Hz
• Eyes: Around 20–90 Hz
• Chest and abdomen: Approximately 60–90 Hz
• Lower limbs: Typically between 2–20 Hz

What is the natural frequency of earth?

#

The Earth’s natural frequency is primarily characterized by the Schumann resonances, a set of extremely low-frequency (ELF) electromagnetic waves that resonate within the cavity formed between the Earth’s surface and the ionosphere. These resonances are excited by lightning discharges and manifest as standing waves encircling the planet.

The fundamental Schumann resonance occurs at approximately 7.83 Hz, often referred to as the Earth’s “heartbeat” or “hum.” This frequency corresponds to a wavelength equal to the Earth’s circumference, resulting from the planet’s spherical geometry and the conductive properties of the ionosphere.

Beyond the fundamental frequency, the Schumann resonances include several higher harmonics, typically observed at: 14.3 Hz, 20.8 Hz, 27.3 Hz, 33.8 Hz

What is special in 432 Hz and 440 Hz?

#

The frequency of 432 Hz has been associated with historical tuning systems. Some proponents suggest that ancient cultures utilized this tuning, believing it to be in harmony with the universe’s natural vibrations. However, the standardization of 440 Hz as the reference pitch (A4) was established in the 20th century, and the historical use of 432 Hz remains a topic of debate.

Listeners often report that music tuned to 432 Hz feels more relaxing and harmonious compared to the standard 440 Hz tuning. Some small-scale studies and anecdotal reports suggest potential benefits, such as reduced stress and anxiety. For instance, a study indicated that 432 Hz music might help decrease dental anxiety during procedures.

How come same frequency have different experience?

#

When we play same musical notes say c# on different instruments say guitar or flute or human sing at c# note they look different. Even two or three guitar all playing c# can be experienced differently by a person of good ears.

When you play the same musical note, like C♯, on different instruments (or with your voice), the sameness comes down to pitch, but the difference is in timbre. Here’s a breakdown of what’s happening:

What Is the “Sameness”?

#

The sameness is the fundamental frequency.

• C♯ has a specific frequency. For example, C♯4 (the C♯ just above middle C) has a frequency of 277.18 Hz.
• Whether you play this on a guitar, flute, or sing it, if you’re accurately producing that frequency, the pitch is perceived as the same.

So, pitch = sameness in frequency = what your brain identifies as the “note.”, not frequency (which physical, existential, finer concept than the note)

🎨 What Is Different Then?

#

The timbre (or tone color) is what makes each instrument sound unique.

Even if the pitch is the same:

• A flute sounds smooth and airy.
• A guitar sounds plucky and resonant.
• A human voice has rich, complex overtones and slight variations in pronunciation and vibration.

This is because of differences in:

• Harmonic content: Each instrument produces a different mix of overtones (frequencies above the fundamental).
• Attack and decay: How the sound starts and fades away.
• Resonance and body shape: Physical structure amplifies different frequencies.
• Noise components: Breath in a flute, string buzz in a guitar, etc.

👂 Why Do We Still Recognize the Note?

#

Your brain can isolate the fundamental frequency among all the overtones and textures, and that’s why you can say, “That’s a C♯,” even if it’s played on very different instruments.

Think of pitch as the word being spoken (e.g., “hello”), and timbre as the voice speaking it. Different people (or instruments) say “hello” in unique voices, but the word remains the same.

Understanding Pitch, Frequency, Amplitude, Loudness

#

Are Pitch and Frequency the Same?

No, they are not exactly the same — but they are closely related.

Frequency : A physical measurement: how many times per second a sound wave vibrates (measured in Hz). Perceived as Objective. Frequency is the input: the vibration of air molecules at a certain rate.

Pitch: The human perception of how “high” or “low” a sound is. Perceived as Subjective. Pitch is the output: what your brain hears and interprets based on that frequency.

• A sound wave at 440 Hz is a frequency.
• You hear it as the musical note A4 — that’s pitch.

Why They’re Not Exactly the Same:

1. Nonlinear Perception: Our ears are more sensitive to certain frequencies (e.g., we hear 1,000–4,000 Hz more easily).
2. Contextual Effects: Two notes of the same frequency might be perceived differently in different musical contexts.
3. Complex Sounds: If a sound contains lots of overtones or noise, the perceived pitch might be influenced by more than just the fundamental frequency.

Musical Example:

• A violin and a trumpet can both play 440 Hz.
• The frequency is the same → they’re playing the same pitch.
• But your brain identifies them as sounding different because of timbre — not pitch.

Analogy:

• Frequency is like the wavelength of light.
• Pitch is like the color you see — similar but dependent on perception.

Fundamental characteristics of sound wave vs sound wave perceived by humans.

#

A Quick Breakdown of ADSR:

These are sub-characteristics of amplitude over time, crucial in music and sound design:

• Attack: How quickly the sound reaches peak amplitude.
• Decay: How fast it drops after the attack.
• Sustain: The level it holds during the main part of the sound.
• Release: How fast it fades after the note ends.

Summary Analogy

What Is an Overtone?

#

Overtones (or harmonics) are integer multiples of a fundamental frequency.

Let’s say:

You play C♯3 on a flute (~138.59 Hz). Then:

1st harmonic (fundamental) = 138.59 Hz → C♯3

2nd harmonic = 277.18 Hz → C♯4

3rd harmonic = 415.77 Hz → G♯4

4th harmonic = 554.36 Hz → C♯5

8th harmonic = 1108.73 Hz → C♯6

How to measure Loudness?

#

Loudness is measured in decibels (dB)

🔊 What is a Decibel (dB)?

#

• The decibel (dB) is a logarithmic unit used to measure the intensity (or pressure) of sound.
• It tells you how much more or less intense a sound is relative to a reference level.

🧠 Loudness vs Sound Pressure Level (SPL):

#

Important Notes:

• 0 dB SPL = the quietest sound a healthy human ear can hear (reference level: 20 µPa).
• 60 dB SPL = typical conversation
• 90 dB SPL = lawn mower
• 120+ dB SPL = rock concert or jet engine — pain threshold
• Every increase of 10 dB roughly doubles the perceived loudness.
  • 60 dB → 70 dB = feels twice as loud
  • 70 dB → 80 dB = again, twice as loud (compared to 70)
• dB SPL: Sound Pressure Level
• dB HL: Hearing Level (used in audiology)
• dB FS: Full Scale (used in digital audio systems)

How come loudness becomes 2 double every 10 db increment?

#

Great questions! Let’s walk through both step by step:

📈 1. Why does loudness “double” every 10 dB?

#

This comes from how human hearing works and how decibels are defined:

Decibels are logarithmic

The decibel scale uses log base 10 — this compresses a wide range of sound intensities into manageable numbers.

The formula to convert a sound intensity level (power) to decibels is:

$$ \text{dB} = 10 \log_{10} \left( \frac{P}{P_0} \right) $$

Where:

• $P$ = measured power
• $P_0$ = reference power (usually the threshold of hearing)

So:

• If you multiply the power of the sound by 10, the decibel level goes up by 10 dB.
• A 10 dB increase in power corresponds to a perceived doubling of loudness (roughly — because loudness is subjective and frequency-dependent).

🧠 Why not exact? Because loudness perception is psychoacoustic — it depends on the ear, frequency, duration, etc.

💨 2. What are the units of pressure? How is it converted into decibels?

#

Sound Pressure Level (SPL)

This is what’s usually measured with microphones:

$$ \text{dB SPL} = 20 \log_{10} \left( \frac{p}{p_0} \right) $$

Where:

• $p$ = measured sound pressure in Pascals (Pa)
• $p_0 = 20 , \mu\text{Pa}$ = reference pressure (the quietest sound a young, healthy human can hear)

Why 20 instead of 10?

Because power is proportional to pressure squared. So if power uses:

$$ \text{dB} = 10 \log_{10}(P / P_0) $$

And power ∝ pressure², then:

$$ \text{dB SPL} = 10 \log_{10}\left(\frac{p^2}{p_0^2}\right) = 20 \log_{10} \left(\frac{p}{p_0}\right) $$

*Example:

Let’s say a sound has pressure amplitude of 0.2 Pa:

$$ \text{dB SPL} = 20 \log_{10} \left( \frac{0.2}{20 \times 10^{-6}} \right) = 20 \log_{10} (10,000) = 20 \times 4 = 80 , \text{dB SPL} $$

So a pressure of 0.2 Pa is 80 dB SPL, which is around busy street traffic or loud conversation.

📝 Summary

#

How much is p0?

#

Reference Pressure $p_0$ in Sound Pressure Level (SPL):

$$ p_0 = 20 , \mu\text{Pa} = 20 \times 10^{-6} , \text{Pa} = 0.00002 , \text{Pa} $$

What does this mean?

• $p_0$ is the threshold of human hearing at 1 kHz, under ideal conditions.
• It’s the quietest sound pressure a healthy young human ear can detect.

For context:

20 µPa (microPascals) is the softest sound pressure that gets mapped to 0 dB SPL — the baseline for sound intensity in human hearing.

Can dB value be negative? what is the meaning of that?

#

Yes, decibel (dB) values can absolutely be negative — and it’s both common and meaningful, especially in acoustics, audio engineering, and signal processing.

A negative dB value means that the measured quantity is less than the reference level.

In sound pressure level (SPL):

$$ \text{dB SPL} = 20 \log_{10} \left( \frac{p}{p_0} \right) $$

If $p < p_0$, then:

• The ratio is less than 1
• The log of a number less than 1 is negative
• Therefore, dB SPL < 0

Example:

Let’s say:

• $p = 2 , \mu\text{Pa} = 2 \times 10^{-6} , \text{Pa}$
• $p_0 = 20 , \mu\text{Pa}$

Then:

$$ \text{dB SPL} = 20 \log_{10} \left( \frac{2 \times 10^{-6}}{20 \times 10^{-6}} \right) = 20 \log_{10} (0.1) = 20 \times (-1) = -20 , \text{dB SPL} $$

✅ So: a sound that is 10× quieter than the threshold of hearing has -20 dB SPL

Can we hear sounds with negative dB SPL?

#

In theory: no, because 0 dB SPL is the threshold of human hearing for a typical young adult in ideal conditions.

But in practice:

• Instruments (like high-end microphones) can detect below 0 dB SPL.
• Negative dB SPL indicates levels that are quieter than human hearing can detect, but still physically measurable.

In other fields:

Summary

What is range of loudness?

#

In theory:

• As $p \to 0$, the ratio $\frac{p}{p_0} \to 0$
• $\log_{10}(0) = -\infty$
• So: theoretical minimum SPL = -∞ dB SPL

But in practice: we hit physical limits.

In Practice: What’s the Quietest Sound Ever Measured?

The world’s quietest place (anechoic chambers):

• Around -20 to -25 dB SPL
• Example: Orfield Labs in Minnesota or Microsoft’s chamber in Washington
• These rooms absorb 99.99% of sound reflections

In these environments, you can hear your heartbeat, blood flow, and joint movements!

Physical Limit: Thermal Noise Floor of Air

• Even in a perfectly silent room, air molecules vibrate randomly due to thermal energy.
• This creates what’s called the Brownian noise or thermal noise floor.

That limit is around:

$$ \text{≈ } -23 , \text{dB SPL (at room temperature)} $$

🔬 So this is about the quietest “real” sound pressure level you can detect in air.

Can we go quieter?

#

Yes — but not in air.

• In outer space, there’s no air, so no sound (sound can’t travel in a vacuum).
• But in low-pressure environments (e.g. deep cryogenic labs), you could measure even lower pressure fluctuations — but not sound in the traditional acoustic sense.

Summary:

What happens to human brain when human is put in -20 dB environment?

#

This kind of silence occurs in anechoic chambers (like at Orfield Labs or Microsoft’s lab). These rooms absorb 99.99% of all sound reflections, and ambient SPLs can go as low as −20 to −23 dB SPL — quieter than anywhere in nature.

What do people experience in these environments?

#

1. Heightened Internal Sounds

You begin to hear:

• Your heartbeat
• Blood rushing through arteries
• Stomach gurgling
• Joints and bones creaking when you move
• Even eye movements can become audible

This can feel unnatural and unsettling, because the brain is not used to hearing itself in this way.

2. Disorientation and Balance Issues

• Humans use echoes and ambient noise subtly to maintain spatial awareness and balance.
• In total silence, your inner ear and vestibular system may feel disconnected.
• Many people lose balance or feel dizzy after a few minutes.

3. Mental Effects

• Extended time in these environments can lead to:

  • Anxiety
  • Sensory deprivation effects
  • Mild hallucinations (especially auditory)
  • Time distortion (minutes can feel like hours)
  • Cognitive strain — the brain searches for input that isn’t there

🧠 The brain needs noise — ambient sound helps it “tune” to the world.

Notes:

• The record for staying in an anechoic chamber is about 45 minutes to 1 hour (by journalists or researchers).
• Most people ask to leave within 15–30 minutes due to discomfort.

Why don’t we get sick in nature where it’s quiet?

#

Even the quietest natural environments (like remote deserts, snowfields, or caves) usually hover around 10–20 dB SPL due to wind, wildlife, or your own movements.

So:

You almost never experience −20 dB SPL naturally — the world always hums at a low level.

The human brain is adapted to a “noisy” world. If you place it in deep unnatural silence, it can become disoriented, anxious, and distressed.

Not “mad” in the medical sense — but mentally and physically uneasy.

Closing Words - Poetic Expression of Silence

#

In this room where sound forgets to live,
Walls do not echo, nor footsteps forgive.
The air hangs thick with unspoken weight,
And time drips slow through a breathless gate.

No whisper dares to float or flee,
Yet every pulse inside of me
Becomes a drum, a beating roar—
Of things I’ve never heard before.

My bones creak softly like distant trees,
My eyes hum gently with unseen seas.
The blood within me writes its song,
And I, the listener, all along.

But stillness grows like creeping vine,
It climbs inside this skull of mine.
Where silence isn’t calm or rest—
It’s pressure mounting in the chest.

And thoughts, untethered, start to spin,
Without the world to anchor in.
I meet myself in echoes lost,
A mirror sharp, a quiet cost.

Then fear arrives—not loud or clear—
But felt in bones, and breath, and ear.
As if the room is not quite real,
And I’m a thought it wants to steal.

I love pin drop silence
But this silence is quite distrubing
Because I am hearing my own blood flow
and blinking of eyes.