my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

my-project-public — gitlawb

#1	import hashlib
#2	import os
#3
#4	import validators
#5
#6	from embedchain.helpers.json_serializable import register_deserializable
#7	from embedchain.loaders.base_loader import BaseLoader
#8
#9	try:
#10	from deepgram import DeepgramClient, PrerecordedOptions
#11	except ImportError:
#12	raise ImportError(
#13	"Audio file requires extra dependencies. Install with `pip install deepgram-sdk==3.2.7`"
#14	) from None
#15
#16
#17	@register_deserializable
#18	class AudioLoader(BaseLoader):
#19	def __init__(self):
#20	if not os.environ.get("DEEPGRAM_API_KEY"):
#21	raise ValueError("DEEPGRAM_API_KEY is not set")
#22
#23	DG_KEY = os.environ.get("DEEPGRAM_API_KEY")
#24	self.client = DeepgramClient(DG_KEY)
#25
#26	def load_data(self, url: str):
#27	"""Load data from a audio file or URL."""
#28
#29	options = PrerecordedOptions(
#30	model="nova-2",
#31	smart_format=True,
#32	)
#33	if validators.url(url):
#34	source = {"url": url}
#35	response = self.client.listen.prerecorded.v("1").transcribe_url(source, options)
#36	else:
#37	with open(url, "rb") as audio:
#38	source = {"buffer": audio}
#39	response = self.client.listen.prerecorded.v("1").transcribe_file(source, options)
#40	content = response["results"]["channels"][0]["alternatives"][0]["transcript"]
#41
#42	doc_id = hashlib.sha256((content + url).encode()).hexdigest()
#43	metadata = {"url": url}
#44
#45	return {
#46	"doc_id": doc_id,
#47	"data": [
#48	{
#49	"content": content,
#50	"meta_data": metadata,
#51	}
#52	],
#53	}
#54

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public