Implementation:Groq Groq python Transcriptions Create

Knowledge Sources	Groq Python SDK Groq API Docs
Domains	Audio, Speech_Recognition
Last Updated	2026-02-15 16:00 GMT

Overview

Concrete tool for submitting audio transcription requests to Groq-hosted Whisper models provided by the Groq Python SDK.

Description

The Transcriptions.create() method sends an audio file to the Groq API endpoint /openai/v1/audio/transcriptions as a multipart form upload. It returns a Transcription object containing the transcribed text. The method supports both file uploads and URL-based audio references.

Usage

Access via client.audio.transcriptions.create(). Provide the audio file and model identifier. Optionally specify language, prompt, response format, and timestamp granularity.

Code Reference

Source Location

• Repository: groq-python
• File: src/groq/resources/audio/transcriptions.py
• Lines: L47-241 (sync), L264-458 (async)

Signature

class Transcriptions(SyncAPIResource):
    def create(
        self,
        *,
        model: Union[str, Literal["whisper-large-v3", "whisper-large-v3-turbo"]],
        file: FileTypes | Omit = omit,
        language: Union[str, Literal["en", "zh", "de", "es", "fr", "ja", ...]] | Omit = omit,
        prompt: str | Omit = omit,
        response_format: Literal["json", "text", "verbose_json"] | Omit = omit,
        temperature: float | Omit = omit,
        timestamp_granularities: List[Literal["word", "segment"]] | Omit = omit,
        url: str | Omit = omit,
    ) -> Transcription:

Import

from groq import Groq
# Access via: client.audio.transcriptions.create(...)

I/O Contract

Inputs

Name	Type	Required	Description
model	str or Literal["whisper-large-v3", "whisper-large-v3-turbo"]	Yes	Whisper model ID
file	FileTypes	No	Audio file (not file name); either file or url required
url	str	No	Audio URL (Base64URL supported); either file or url required
language	str	No	ISO-639-1 language code for improved accuracy
prompt	str	No	Optional text to guide transcription style
response_format	Literal["json", "text", "verbose_json"]	No	Output format
temperature	float	No	Sampling temperature for transcription
timestamp_granularities	List[Literal["word", "segment"]]	No	Timestamp detail level

Outputs

Name	Type	Description
(return)	Transcription	Object with .text field containing the transcribed text

Usage Examples

Basic Transcription

from groq import Groq

client = Groq()

with open("interview.mp3", "rb") as audio_file:
    transcription = client.audio.transcriptions.create(
        file=audio_file,
        model="whisper-large-v3",
    )

print(transcription.text)

With Language and Prompt

transcription = client.audio.transcriptions.create(
    file=open("meeting.wav", "rb"),
    model="whisper-large-v3-turbo",
    language="en",
    prompt="Technical discussion about machine learning and LLMs",
    response_format="verbose_json",
    timestamp_granularities=["segment"],
)

Related Pages

Implements Principle

• Principle:Groq_Groq_python_Audio_Transcription_Request

Requires Environment

• Environment:Groq_Groq_python_Python_Groq_SDK