Phase 1: Audio Processing & Streaming · 50 min · Python · PyAudio · pydub

Digital Audio Basics: Sample Rate, Bit Depth, and Channels

Before you stream audio, you must understand what audio is at the sample level.

Hiring signal: Digital audio manipulation is the first coding round in streaming audio interviews at Deepgram and ElevenLabs.

What you will learn

Read and write WAV files with Python's wave module and pydub
Manipulate sample rate, bit depth, and channels programmatically
Visualize audio waveforms and spectrograms with matplotlib
Understand the relationship between sample rate, bit depth, and audio quality

The Problem

Every voice agent pipeline starts with audio I/O. You need to read audio from telephony (8kHz μ-law), convert it to the format ASR expects (16kHz PCM), process it, and write TTS output (24kHz PCM). If you can't reliably read, write, and inspect audio files, nothing downstream works.

This lesson builds the audio I/O muscle: read WAV files, understand their properties, generate test audio, and visualize what's inside.

The Concept

The WAV File Format

WAV is the standard uncompressed audio format. It's a RIFF container with a header followed by raw PCM samples:

RIFF Header (12 bytes)
  - "RIFF" (4 bytes)
  - File size - 8 (4 bytes)
  - "WAVE" (4 bytes)

fmt Chunk (24+ bytes)
  - "fmt " (4 bytes)
  - Chunk size (4 bytes)
  - Audio format: 1=PCM, 7=μ-law (2 bytes)
  - Num channels (2 bytes)
  - Sample rate (4 bytes)
  - Byte rate (4 bytes)
  - Block align (2 bytes)
  - Bits per sample (2 bytes)

data Chunk
  - "data" (4 bytes)
  - Data size (4 bytes)
  - Raw audio samples

A WAV file has sample_rate=16000, bits_per_sample=16, channels=1. What is the byte rate?

16000 × 16 = 256000 bytes/sec

Reading WAV Files with Python

import wave
import numpy as np

def read_wav(filepath):
    with wave.open(filepath, 'rb') as wf:
        sample_rate = wf.getframerate()
        channels = wf.getnchannels()
        sample_width = wf.getsampwidth()
        n_frames = wf.getnframes()
        raw_data = wf.readframes(n_frames)

    # Convert bytes to numpy array
    if sample_width == 2:  # 16-bit
        audio = np.frombuffer(raw_data, dtype=np.int16)
    elif sample_width == 1:  # 8-bit
        audio = np.frombuffer(raw_data, dtype=np.uint8)
    elif sample_width == 4:  # 32-bit
        audio = np.frombuffer(raw_data, dtype=np.int32)

    if channels > 1:
        audio = audio.reshape(-1, channels)

    return audio, sample_rate, channels, sample_width

Writing WAV Files

def write_wav(filepath, audio, sample_rate, bit_depth=16):
    with wave.open(filepath, 'wb') as wf:
        wf.setnchannels(1)
        wf.setsampwidth(bit_depth // 8)
        wf.setframerate(sample_rate)
        wf.writeframes(audio.tobytes())

Visualizing Audio

Unlock the full lesson

You've read the first 2 sections. The rest of this lesson covers Build It, Use It, Ship It, Evaluation, Key Terms, Common Pitfalls, Interview Framing — plus a hands-on lab, quiz, and project artifact.

Create a free account to unlock Phase 0 and Phase 1 of every course — no credit card.