Agno agents can transcribe audio files using different tools and models. You can use native capabilities of OpenAI or fully multimodal Gemini models.
Examples of ways to do Audio to Text (Transcribe) are:

​
Using OpenAI Whisper (Cloud)

The following agent uses OpenAI Whisper API for audio transcription.
cookbook/tools/models/openai_tools.py
import base64
from pathlib import Path

from agno.agent import Agent
from agno.run.agent import RunOutput
from agno.tools.openai import OpenAITools
from agno.utils.media import download_file, save_base64_data

# Example 1: Transcription
url = "https://agno-public.s3.amazonaws.com/demo_data/sample_conversation.wav"

local_audio_path = Path("tmp/sample_conversation.wav")
print(f"Downloading file to local path: {local_audio_path}")
download_file(url, local_audio_path)

transcription_agent = Agent(
    tools=[OpenAITools(transcription_model="gpt-4o-transcribe")],
    markdown=True,
)
transcription_agent.print_response(
    f"Transcribe the audio file for this file: {local_audio_path}"
)
Best for: High accuracy, cloud processing

​
Using Multimodal Models

Multimodal models like Gemini can transcribe audio directly without additional tools.
cookbook/agents/multimodal/audio_to_text.py
import requests
from agno.agent import Agent
from agno.media import Audio
from agno.models.google import Gemini

agent = Agent(
    model=Gemini(id="gemini-2.0-flash-exp"),
    markdown=True,
)

url = "https://agno-public.s3.us-east-1.amazonaws.com/demo_data/QA-01.mp3"

response = requests.get(url)
audio_content = response.content

# Give a transcript of this audio conversation. Use speaker A, speaker B to identify speakers.

agent.print_response(
    "Give a transcript of this audio conversation. Use speaker A, speaker B to identify speakers.",
    audio=[Audio(content=audio_content)],
    stream=True,
)
Best for: Direct model integration, conversation understanding

​
Team-Based Transcription

Teams can handle complex audio processing workflows with multiple specialized agents.
cookbook/teams/multimodal/audio_to_text.py
import requests
from agno.agent import Agent
from agno.media import Audio
from agno.models.google import Gemini
from agno.team import Team

transcription_specialist = Agent(
    name="Transcription Specialist",
    role="Convert audio to accurate text transcriptions",
    model=Gemini(id="gemini-2.0-flash-exp"),
    instructions=[
        "Transcribe audio with high accuracy",
        "Identify speakers clearly as Speaker A, Speaker B, etc.",
        "Maintain conversation flow and context",
    ],
)

content_analyzer = Agent(
    name="Content Analyzer",
    role="Analyze transcribed content for insights",
    model=Gemini(id="gemini-2.0-flash-exp"),
    instructions=[
        "Analyze transcription for key themes and insights",
        "Provide summaries and extract important information",
    ],
)

# Create a team for collaborative audio-to-text processing
audio_team = Team(
    name="Audio Analysis Team",
    model=Gemini(id="gemini-2.0-flash-exp"),
    members=[transcription_specialist, content_analyzer],
    instructions=[
        "Work together to transcribe and analyze audio content.",
        "Transcription Specialist: First convert audio to accurate text with speaker identification.",
        "Content Analyzer: Analyze transcription for insights and key themes.",
    ],
    markdown=True,
)

url = "https://agno-public.s3.us-east-1.amazonaws.com/demo_data/QA-01.mp3"

response = requests.get(url)
audio_content = response.content

audio_team.print_response(
    "Give a transcript of this audio conversation. Use speaker A, speaker B to identify speakers.",
    audio=[Audio(content=audio_content)],
    stream=True,
)
Best for: Complex workflows, multiple processing steps

​
Developer Resources