Skip to main content
Transcribe audio files into text using OpenAI’s Whisper model. Supports multiple audio formats and languages.

​
Basic usage

Transcribe an audio file to text: 
import os
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus(
    api_key=os.environ.get("DEDALUS_API_KEY")
)

transcription = client.audio.transcriptions.create(
    file=Path("/path/to/audio.mp3"),
    model="openai/whisper-1"
)

print(transcription.text)

​
Async usage

Transcribe audio asynchronously: 
import os
import asyncio
from pathlib import Path
from dedalus_labs import AsyncDedalus

client = AsyncDedalus(
    api_key=os.environ.get("DEDALUS_API_KEY")
)

async def main():
    transcription = await client.audio.transcriptions.create(
        file=Path("/path/to/audio.mp3"),
        model="openai/whisper-1"
    )
    print(transcription.text)

asyncio.run(main())

​
Supported formats

The transcription API supports the following audio formats:
  • mp3
  • mp4
  • mpeg
  • mpga
  • m4a
  • wav
  • webm
Maximum file size: 25 MB

​
File upload methods

Use Path objects for local files:
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

transcription = client.audio.transcriptions.create(
    file=Path("/path/to/audio.mp3"),
    model="openai/whisper-1"
)

​
Language specification

Improve accuracy by specifying the input language using ISO-639-1 codes: 
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

# Transcribe Spanish audio
transcription = client.audio.transcriptions.create(
    file=Path("/path/to/spanish_audio.mp3"),
    model="openai/whisper-1",
    language="es"  # ISO-639-1 code for Spanish
)

print(transcription.text)
  • en - English
  • es - Spanish
  • fr - French
  • de - German
  • it - Italian
  • pt - Portuguese
  • nl - Dutch
  • ja - Japanese
  • ko - Korean
  • zh - Chinese
  • ru - Russian
  • ar - Arabic
  • hi - Hindi

​
Response formats

Choose from multiple output formats: 
transcription = client.audio.transcriptions.create(
    file=Path("/path/to/audio.mp3"),
    model="openai/whisper-1",
    response_format="json"  # Default
)

print(transcription.text)

​
Prompting for context

Provide context to guide the transcription style and content: 
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

# Guide the model with context about the audio
transcription = client.audio.transcriptions.create(
    file=Path("/path/to/audio.mp3"),
    model="openai/whisper-1",
    prompt="This is a technical interview about machine learning and neural networks."
)

print(transcription.text)
The prompt parameter helps the model:
  • Maintain consistent spelling of uncommon words
  • Match a specific writing style
  • Continue from previous audio segments
  • Use domain-specific terminology correctly

​
Temperature control

Adjust the sampling temperature for different transcription behaviors: 
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

# Lower temperature for more consistent transcriptions
transcription = client.audio.transcriptions.create(
    file=Path("/path/to/audio.mp3"),
    model="openai/whisper-1",
    temperature=0.0  # Range: 0.0 to 1.0
)

print(transcription.text)
  • temperature=0.0 - More deterministic and consistent
  • temperature=0.5 - Balanced (default)
  • temperature=1.0 - More varied transcriptions

​
Complete example

Transcribe with all options: 
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

transcription = client.audio.transcriptions.create(
    file=Path("/path/to/interview.mp3"),
    model="openai/whisper-1",
    language="en",
    prompt="This is an interview with Dr. Smith about quantum computing.",
    response_format="verbose_json",
    temperature=0.2
)

print(f"Language: {transcription.language}")
print(f"Duration: {transcription.duration}s")
print(f"\nFull text:\n{transcription.text}")

print("\nSegments:")
for i, segment in enumerate(transcription.segments, 1):
    print(f"{i}. [{segment.start:.2f}s - {segment.end:.2f}s]: {segment.text}")

​
Error handling

import dedalus_labs
from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

try:
    transcription = client.audio.transcriptions.create(
        file=Path("/path/to/audio.mp3"),
        model="openai/whisper-1"
    )
    print(transcription.text)
except FileNotFoundError:
    print("Audio file not found")
except dedalus_labs.BadRequestError as e:
    print(f"Invalid request: {e.message}")
    # Common causes: file too large, unsupported format
except dedalus_labs.APIConnectionError as e:
    print("Network error occurred")
    print(e.__cause__)
except dedalus_labs.APIStatusError as e:
    print(f"API error: {e.status_code}")

​
Processing multiple files

from pathlib import Path
from dedalus_labs import Dedalus

client = Dedalus()

audio_dir = Path("/path/to/audio_files")
audio_files = list(audio_dir.glob("*.mp3"))

for audio_file in audio_files:
    transcription = client.audio.transcriptions.create(
        file=audio_file,
        model="openai/whisper-1",
        language="en"
    )
    
    # Save transcription to text file
    output_file = audio_file.with_suffix(".txt")
    output_file.write_text(transcription.text)
    print(f"Transcribed: {audio_file.name}")
Ensure your audio files are under 25 MB. For larger files, consider:
  • Compressing the audio
  • Splitting into smaller segments
  • Using a lower bitrate format

Build docs developers (and LLMs) love

Get started for freeTalk to us