openai.Audio.transcribe() fails with BytesIO unless .name is manually set

[cyborg728]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

When using openai_client.audio.transcriptions.create) with an in-memory file (io.BytesIO), the function fails unless the .name attribute is manually set to a valid filename with a supported extension (e.g., audio.wav or audio.mp3).

This requirement is not documented, and the resulting error is confusing (e.g., AttributeError). This is especially problematic when handling file uploads in memory, such as from Telegram bots or web clients.

To Reproduce

import io
from openai import AsyncOpenAI

openai_client = AsyncOpenAI(api_key=OPENAI_API_KEY)

# Simulating an audio file from memory (e.g., from a Telegram bot)

audio_bytes = io.BytesIO(b"audio binary data here")

# This will fail unless you set `audio_bytes.name = "audio.wav"

await openai_client.audio.transcriptions.create(
        model="whisper-1", file=audio_bytes
)

Code snippets

OS

windows

Python version

python v3.11.4

Library version

openai v1.75.0

[issamhaimour]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

When using openai_client.audio.transcriptions.create) with an in-memory file (io.BytesIO), the function fails unless the .name attribute is manually set to a valid filename with a supported extension (e.g., audio.wav or audio.mp3).

This requirement is not documented, and the resulting error is confusing (e.g., AttributeError). This is especially problematic when handling file uploads in memory, such as from Telegram bots or web clients.

To Reproduce

import io
from openai import AsyncOpenAI

openai_client = AsyncOpenAI(api_key=OPENAI_API_KEY)

# Simulating an audio file from memory (e.g., from a Telegram bot)

audio_bytes = io.BytesIO(b"audio binary data here")

# This will fail unless you set `audio_bytes.name = "audio.wav"

await openai_client.audio.transcriptions.create(
        model="whisper-1", file=audio_bytes
)

Code snippets

OS

windows

Python version

python v3.11.4

Library version

openai v1.75.0