File size: 2,653 Bytes
31defb6 979d3c4 31defb6 979d3c4 b10969b 5b6b68a e47af5b adbb180 15134ce 31defb6 96369f4 63a0440 31defb6 a2934a4 08a4021 61a2310 a11f33b 3bf7ebf 6a7b968 31defb6 5f148df 31defb6 08fa659 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
import whisper
import os
from pytube import YouTube
import pandas as pd
import plotly_express as px
import nltk
import plotly.graph_objects as go
from optimum.onnxruntime import ORTModelForSequenceClassification
from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification, AutoModelForTokenClassification
from sentence_transformers import SentenceTransformer, CrossEncoder, util
import streamlit as st
import en_core_web_lg
from functions import *
nltk.download('punkt')
from nltk import sent_tokenize
st.sidebar.header("Home")
asr_model_options = ['base.en','small.en']
asr_model_name = st.sidebar.selectbox("Whisper Model Options", options=asr_model_options, key='sbox')
with st.spinner('Loading all models...'):
sent_pipe, sum_pipe, ner_pipe, cross_encoder, kg_model, kg_tokenizer, emb_tokenizer = load_models()
st.markdown("## Earnings Call Analysis Whisperer")
st.markdown(
"""
This app assists finance analysts with transcribing and analysis Earnings Calls by carrying out the following tasks:
- Transcribing earnings calls using Open AI's [Whisper](https://github.com/openai/whisper).
- Analysing the sentiment of transcribed text using the quantized version of [FinBert-Tone](https://huggingface.co/nickmuchi/quantized-optimum-finbert-tone).
- Summarization of the call with [FaceBook-Bart-Large-CNN](https://huggingface.co/facebook/bart-large-cnn) model with entity extraction
- Question Answering engine powered by Langchain and [Sentence Transformers](https://huggingface.co/sentence-transformers/all-mpnet-base-v2).
- Knowledge Graph generation using [Babelscape/rebel-large](https://huggingface.co/Babelscape/rebel-large) model.
**👇 Enter a YouTube Earnings Call URL below and navigate to the sidebar tabs**
"""
)
if 'sbox' not in st.session_state:
st.session_state.sbox = asr_model_name
if "url" not in st.session_state:
st.session_state.url = "https://www.youtube.com/watch?v=iWobmXvCM0c"
if "earnings_passages" not in st.session_state:
st.session_state["earnings_passages"] = ''
if "sen_df" not in st.session_state:
st.session_state['sen_df'] = ''
url_input = st.text_input(
label="Enter YouTube URL, example below is FaceBook Earnings Call 2022", key="url")
st.markdown(
"<h3 style='text-align: center; color: red;'>OR</h3>",
unsafe_allow_html=True
)
upload_wav = st.file_uploader("Upload a .wav sound file ",key="upload")
auth_token = os.environ.get("auth_token")
st.markdown("")
|