changed task creation to server

app.py

@@ -14,12 +14,15 @@ import pandas as pd
 import requests
 import streamlit as st
 from audio_recorder_streamlit import audio_recorder
+import torchaudio
 
 from logger import logger
 from utils import fs
-from enums import SAVE_PATH, ELO_JSON_PATH, ELO_CSV_PATH, EMAIL_PATH, TEMP_DIR, CREATE_TASK_URL,TRANSCRIBE_URL
+from enums import SAVE_PATH, ELO_JSON_PATH, ELO_CSV_PATH, EMAIL_PATH, TEMP_DIR, NEW_TASK_URL,ARENA_PATH
 
 result_queue = Queue()
+random_df = pd.read_csv("random_audios.csv")
+random_paths = random_df["path"].tolist()
 
 def result_writer_thread():
     result_writer = ResultWriter(SAVE_PATH)
@@ -137,9 +140,9 @@ def send_task(payload):
         "Authorization": f"Bearer {os.getenv('CREATE_TASK_API_KEY')}"
     }
     if payload["task"] in ["fetch_audio","write_result"]:
-        response = requests.post(CREATE_TASK_URL,json=payload,headers=header,timeout=300)
+        response = requests.post(NEW_TASK_URL,json=payload,headers=header,timeout=300)
     else:
-        response = requests.post(TRANSCRIBE_URL,json=payload,headers=header,timeout=300,stream=True)
+        response = requests.post(NEW_TASK_URL,json=payload,headers=header,timeout=300,stream=True)
     try:
         response = response.json()
     except Exception as e:
@@ -152,12 +155,12 @@ def send_task(payload):
     if payload["task"] == "transcribe_with_fastapi":
         return response["text"]
 
-    elif payload["task"] == "fetch_audio":
-        array = response["array"]
-        array = decode_audio_array(array)
-        sampling_rate = response["sample_rate"]
-        filepath = response["filepath"]
-        return array,sampling_rate,filepath
+def fetch_audio():
+    filepath = random.choice(random_paths)
+    with fs.open(f"{ARENA_PATH}/{filepath}", 'rb') as f:
+        audio,sr = torchaudio.load(f)
+        audio = audio.numpy()
+        return audio,sr,filepath
 
 def encode_audio_array(audio_array):
     buffer = io.BytesIO()
@@ -184,7 +187,7 @@ def call_function(model_name):
         sr = st.session_state.audio['sample_rate']
         array = st.session_state.audio['data']
         if sr != 22050:
-            array = librosa.resample(array,sr,22050)
+            array = librosa.resample(y=array,orig_sr=sr,target_sr=22050)
         encoded_array = encode_audio_array(array)
         payload = {
                 "task":"transcribe_with_fastapi",
@@ -251,6 +254,7 @@ def on_option_1_click():
                 }
             )
             st.session_state.option_selected = True
+            st.session_state.disable_voting=True
 
 def on_option_2_click():
     if st.session_state.transcribed and not st.session_state.option_selected:
@@ -269,6 +273,7 @@ def on_option_2_click():
                 }
             )
             st.session_state.option_selected = True
+            st.session_state.disable_voting=True
 
 def on_option_both_click():
     if st.session_state.transcribed and not st.session_state.option_selected:
@@ -288,6 +293,7 @@ def on_option_both_click():
                 }
             )
             st.session_state.option_selected = True
+            st.session_state.disable_voting=True
 
 def on_option_none_click():
     if st.session_state.transcribed and not st.session_state.option_selected:
@@ -306,6 +312,7 @@ def on_option_none_click():
                 }
             )
             st.session_state.option_selected = True
+            st.session_state.disable_voting=True
 
 def on_click_transcribe():
     if st.session_state.has_audio:
@@ -318,12 +325,13 @@ def on_click_transcribe():
             st.session_state.option_1_model_name_state = ""
             st.session_state.option_2_model_name_state = ""
             st.session_state.option_selected = None
+            st.session_state.recording=True
+            st.session_state.disable_voting=False
 
 def on_random_click():
     reset_state()
     with st.spinner("Fetching random audio... please wait"):
-        fetch_audio_payload = {"task": "fetch_audio"}
-        array, sampling_rate, filepath = send_task(fetch_audio_payload)
+        array, sampling_rate, filepath = fetch_audio()
         st.session_state.audio = {"data":array,"sample_rate":sampling_rate,"format":"audio/wav"}
         st.session_state.has_audio = True
         st.session_state.current_audio_type = "random"
@@ -364,7 +372,8 @@ def main():
         st.session_state.user_email = ""
     if "recording" not in st.session_state:
         st.session_state.recording = True
-
+    if "disable_voting" not in st.session_state:
+        st.session_state.disable_voting = True
     col1, col2 = st.columns([1, 1])
 
     with col1:
@@ -424,16 +433,16 @@ def main():
     c1, c2, c3, c4 = st.columns(4)
 
     with c1:
-        st.button("Prefer Option 1",on_click=on_option_1_click,key="option1_btn")
+        st.button("Prefer Option 1",on_click=on_option_1_click,key="option1_btn",disabled=st.session_state.disable_voting)
 
     with c2:
-        st.button("Prefer Option 2",on_click=on_option_2_click,key="option2_btn")
+        st.button("Prefer Option 2",on_click=on_option_2_click,key="option2_btn",disabled=st.session_state.disable_voting)
 
     with c3:
-        st.button("Prefer Both",on_click=on_option_both_click,key="both_btn")
+        st.button("Prefer Both",on_click=on_option_both_click,key="both_btn",disabled=st.session_state.disable_voting)
 
     with c4:
-        st.button("Prefer None",on_click=on_option_none_click,key="none_btn")
+        st.button("Prefer None",on_click=on_option_none_click,key="none_btn",disabled=st.session_state.disable_voting)
 
     with st.container():
         st.button("New Match",on_click=on_reset_click,key="reset_btn",use_container_width=True)

enums.py

@@ -5,5 +5,5 @@ ELO_JSON_PATH = f"s3://{os.getenv('AWS_BUCKET_NAME')}/{os.getenv('ELO_JSON_PATH'
 ELO_CSV_PATH = f"s3://{os.getenv('AWS_BUCKET_NAME')}/{os.getenv('ELO_CSV_KEY')}"
 EMAIL_PATH = f"s3://{os.getenv('AWS_BUCKET_NAME')}/{os.getenv('EMAILS_KEY')}"
 TEMP_DIR = f"s3://{os.getenv('AWS_BUCKET_NAME')}/{os.getenv('AUDIOS_KEY')}"
-CREATE_TASK_URL = os.getenv("CREATE_TASK_URL")
-TRANSCRIBE_URL = os.getenv("TRANSCRIPTION_TASK_URL")
+NEW_TASK_URL=os.getenv("NEW_TASK_URL")
+ARENA_PATH=os.getenv('ARENA_PATH')

requirements.txt

@@ -8,4 +8,5 @@ scipy
 streamlit==1.40.2
 fsspec==2024.10.0
 boto3
-s3fs
+s3fs
+torchaudio