mirror of
https://github.com/QuivrHQ/quivr.git
synced 2024-12-19 20:31:50 +03:00
9db8b63bb9
This pull request adds a new optional field "tags" to the AssistantOutput classes. The "tags" field allows for categorizing the outputs and can be used for filtering or organizing purposes.
89 lines
2.5 KiB
Python
89 lines
2.5 KiB
Python
import os
|
|
from tempfile import NamedTemporaryFile
|
|
|
|
from logger import get_logger
|
|
from modules.assistant.dto.outputs import (
|
|
AssistantOutput,
|
|
InputFile,
|
|
Inputs,
|
|
OutputBrain,
|
|
OutputEmail,
|
|
Outputs,
|
|
)
|
|
from modules.assistant.ito.ito import ITO
|
|
from openai import OpenAI
|
|
|
|
logger = get_logger(__name__)
|
|
|
|
|
|
class AudioTranscriptAssistant(ITO):
|
|
|
|
def __init__(
|
|
self,
|
|
**kwargs,
|
|
):
|
|
super().__init__(
|
|
**kwargs,
|
|
)
|
|
|
|
async def process_assistant(self):
|
|
client = OpenAI()
|
|
|
|
logger.info(f"Processing audio file {self.uploadFile.filename}")
|
|
|
|
# Extract the original filename and create a temporary file with the same name
|
|
filename = os.path.basename(self.uploadFile.filename)
|
|
temp_file = NamedTemporaryFile(delete=False, suffix=filename)
|
|
|
|
# Write the uploaded file's data to the temporary file
|
|
data = await self.uploadFile.read()
|
|
temp_file.write(data)
|
|
temp_file.close()
|
|
|
|
# Open the temporary file and pass it to the OpenAI API
|
|
with open(temp_file.name, "rb") as file:
|
|
transcription = client.audio.transcriptions.create(
|
|
model="whisper-1", file=file, response_format="text"
|
|
)
|
|
logger.info(f"Transcription: {transcription}")
|
|
|
|
# Delete the temporary file
|
|
os.remove(temp_file.name)
|
|
|
|
return await self.create_and_upload_processed_file(
|
|
transcription, self.uploadFile.filename, "Audio Transcript"
|
|
)
|
|
|
|
|
|
def audio_transcript_inputs():
|
|
output = AssistantOutput(
|
|
name="Audio Transcript",
|
|
description="Transcribes an audio file",
|
|
tags=["new"],
|
|
input_description="One audio file to transcribe",
|
|
output_description="Transcription of the audio file",
|
|
inputs=Inputs(
|
|
files=[
|
|
InputFile(
|
|
key="audio_file",
|
|
allowed_extensions=["mp3", "wav", "ogg", "m4a"],
|
|
required=True,
|
|
description="The audio file to transcribe",
|
|
)
|
|
]
|
|
),
|
|
outputs=Outputs(
|
|
brain=OutputBrain(
|
|
required=True,
|
|
description="The brain to which to upload the document",
|
|
type="uuid",
|
|
),
|
|
email=OutputEmail(
|
|
required=True,
|
|
description="Send the document by email",
|
|
type="str",
|
|
),
|
|
),
|
|
)
|
|
return output
|