AudioContent

AudioContent

The Audio class handles audio data for multimodal AI interactions.

Import

from pyai.multimodal import Audio

Creating Audio

From File

audio = Audio.from_file("recording.mp3")

From URL

audio = Audio.from_url("https://example.com/audio.wav")

From Bytes

with open("audio.wav", "rb") as f:
    audio = Audio.from_bytes(f.read(), format="wav")

From NumPy Array

import numpy as np

# Generate audio data
samples = np.sin(2 * np.pi * 440 * np.linspace(0, 1, 44100))
audio = Audio.from_numpy(samples, sample_rate=44100)

Properties

Property	Type	Description
duration	float	Duration in seconds
sample_rate	int	Samples per second
channels	int	Number of channels
format	str	Audio format
size_bytes	int	File size

Methods

convert()

Convert to different format:

# Convert to MP3
mp3_audio = audio.convert(format="mp3", bitrate=192)

# Convert to WAV
wav_audio = audio.convert(format="wav")

# Convert sample rate
resampled = audio.convert(sample_rate=16000)

trim()

Trim audio:

# Trim to specific duration
trimmed = audio.trim(start=0.0, end=30.0)  # First 30 seconds

# Trim from start
trimmed = audio.trim(start=5.0)  # Skip first 5 seconds

split()

Split into segments:

# Split into 30-second chunks
segments = audio.split(segment_duration=30.0)

save()

Save to file:

audio.save("output.mp3")
audio.save("output.wav", format="wav")

transcribe()

Transcribe audio to text:

text = audio.transcribe()
print(text)

# With options
result = audio.transcribe(
    language="en",
    timestamps=True
)

Using with Agents

With Multimodal Content

from pyai.multimodal import MultimodalContent, Audio

content = MultimodalContent()
content.add_text("Please transcribe and summarize this audio:")
content.add_audio(Audio.from_file("meeting.mp3"))

response = agent.run(content)

Audio Analysis

from pyai import ask
from pyai.multimodal import Audio

audio = Audio.from_file("speech.wav")
response = ask(
    "What language is being spoken and what is the topic?",
    audio=[audio]
)

Format Support

Format	Read	Write	Notes
MP3	✅	✅	Most common
WAV	✅	✅	Uncompressed
M4A	✅	✅	AAC encoded
FLAC	✅	✅	Lossless
OGG	✅	✅	Open format
WebM	✅	✅	Web optimized

Audio Processing

Volume Adjustment

# Increase volume
louder = audio.adjust_volume(gain_db=3.0)

# Decrease volume
quieter = audio.adjust_volume(gain_db=-3.0)

# Normalize
normalized = audio.normalize()

Channel Operations

# Convert to mono
mono = audio.to_mono()

# Convert to stereo
stereo = audio.to_stereo()

# Extract channel
left_channel = audio.get_channel(0)

See Also

• Multimodal-Module - Module overview
• ImageContent - Image handling
• VideoContent - Video handling
• Voice-Module - Real-time voice