Update app.py

app.py

@@ -27,6 +27,12 @@ import os
 from rdkit import RDConfig
 from rdkit.Chem.Features.ShowFeats import _featColors as featColors
 from rdkit.Chem.FeatMaps import FeatMaps
+from elevenlabs.client import ElevenLabs
+from elevenlabs import stream
+
+eleven_key = os.getenv("eleven_key")
+
+elevenlabs = ElevenLabs(api_key=eleven_key)
 
 fdef = AllChem.BuildFeatureFactory(os.path.join(RDConfig.RDDataDir,'BaseFeatures.fdef'))
 
@@ -357,8 +363,10 @@ def parser_node(state: State) -> State:
   props_string = state["props_string"]
   query_task = state["query_task"]
 
-prompt = f'Using only the information provided in the CONTEXT below, answer the QUERY_TASK.\n Your answer musr:\n Directly address the QUERY_TASK.\n Use only facts found in the CONTEXT (do not invent information).\n Be concise, precise and logically consistent.\n End your answer with a "#"'
-QUERY_TASK: {query_task}.\n
+prompt = f'Using only the information provided in the CONTEXT below, answer the QUERY_TASK.\n \
+Your answer must:\n Directly address the QUERY_TASK.\n Use only facts found in the CONTEXT \
+(do not invent information).\n Be concise, precise and logically consistent.\n End your answer with a "#" \
+QUERY_TASK: {query_task}.\n \
 CONTEXT: {props_string}.\n '
 
   res = chat_model.invoke(prompt)
@@ -472,7 +480,28 @@ def PropAgent(task, smiles, reference):
   else:
     img = Image.new('RGB', (250, 250), color = (255, 255, 255))
 
-  return replies[-1], img
+  elita_text = replies[-1]
+
+  voice_settings = {
+          "stability": 0.37,
+          "similarity_boost": 0.90,
+          "style": 0.0,
+          "speed": 1.05
+      }
+
+  audio_stream = elevenlabs.text_to_speech.convert(
+      text = elita_text,
+      voice_id = 'G5KS88IIzHIX1ogRxdrA',
+      model_id = 'eleven_multilingual_v2',
+      output_format='mp3_44100_128',
+      voice_settings=voice_settings
+  )
+
+  audio_converted = b"".join(audio_stream)
+  audio = base64.b64encode(audio_converted).decode("utf-8")
+  audio_player = f'<audio src="data:audio/mpeg;base64,{audio}" controls autoplay></audio>'
+
+  return replies[-1], img, audio_player
 
 with gr.Blocks(fill_height=True) as forest:
   gr.Markdown('''
@@ -492,9 +521,10 @@ with gr.Blocks(fill_height=True) as forest:
     with gr.Column():
       props = gr.Textbox(label="Agent results: ", lines=20 )
       pic = gr.Image(label="Molecule")
+    voice = gr.HTML()
 
 
-      calc_btn.click(PropAgent, inputs = [task, smiles, ref], outputs = [props, pic])
-      task.submit(PropAgent, inputs = [task, smiles, ref], outputs = [props, pic])
+      calc_btn.click(PropAgent, inputs = [task, smiles, ref], outputs = [props, pic, voice])
+      task.submit(PropAgent, inputs = [task, smiles, ref], outputs = [props, pic, voice])
 
 forest.launch(debug=False, mcp_server=True)