add whisper

app.py

@@ -2,16 +2,70 @@ import os
 import gradio as gr
 import requests
 import inspect
+import openai
 import pandas as pd
-from smolagents import OpenAIServerModel, DuckDuckGoSearchTool, PythonInterpreterTool, CodeAgent, WikipediaSearchTool, \
-    SpeechToTextTool
+from smolagents import OpenAIServerModel, DuckDuckGoSearchTool, PythonInterpreterTool, CodeAgent, WikipediaSearchTool
 from pathlib import Path
 import tempfile
+from smolagents.tools import PipelineTool
 
 # (Keep Constants as is)
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
+class SpeechToTextTool(PipelineTool):
+    """
+    Transcribes an audio file to text using the OpenAI Whisper API.
+    Only local file paths are supported.
+    """
+    default_checkpoint = "openai/whisper-1"   # purely informational here
+    description = (
+        "This tool sends an audio file to OpenAI Whisper and returns the "
+        "transcribed text."
+    )
+    name = "transcriber"
+    inputs = {
+        "audio": {
+            "type": "string",
+            "description": "Absolute or relative path to a local audio file.",
+        }
+    }
+    output_type = "string"
+
+    # ──────────────────────────────────────────────────────────────────────────
+    # Public interface
+    # ──────────────────────────────────────────────────────────────────────────
+    def __call__(self, audio: str) -> str:
+        """
+        Convenience wrapper so the tool can be used like a regular function:
+            text = SpeechToTextTool()(path_to_audio)
+        """
+        return self._transcribe(audio)
+
+    # ──────────────────────────────────────────────────────────────────────────
+    # Internal helpers
+    # ──────────────────────────────────────────────────────────────────────────
+    @staticmethod
+    def _transcribe(audio_path: str) -> str:
+        # ----- validation ----------------------------------------------------
+        if not isinstance(audio_path, str):
+            raise TypeError(
+                "Parameter 'audio' must be a string containing the file path."
+            )
+        path = Path(audio_path).expanduser().resolve()
+        if not path.is_file():
+            raise FileNotFoundError(f"No such audio file: {path}")
+
+        # ----- API call ------------------------------------------------------
+        with path.open("rb") as fp:
+            response = openai.audio.transcriptions.create(
+                file=fp,
+                model="whisper-1",        # currently the only Whisper model
+                response_format="text"    # returns plain text instead of JSON
+            )
+
+        # For response_format="text", `response` is already the raw transcript
+        return response
 
 def download_file_if_any(base_api_url: str, task_id: str) -> str | None:
     """
@@ -55,7 +109,7 @@ class BasicAgent:
             model=OpenAIServerModel(model_id="gpt-4o"),
             tools=[DuckDuckGoSearchTool(), WikipediaSearchTool(), SpeechToTextTool()],
             add_base_tools=True,
-            additional_authorized_imports=['pandas','numpy','csv']
+            additional_authorized_imports=['pandas','numpy','csv','subprocess']
         )
 
         print("BasicAgent initialized.")
@@ -72,7 +126,7 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
     and displays the results.
     """
     # --- Determine HF Space Runtime URL and Repo URL ---
-    space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
+    space_id = "l3xv/Final_Assignment_Template"
 
     if profile:
         username= f"{profile.username}"
@@ -242,7 +296,7 @@ if __name__ == "__main__":
     print("\n" + "-"*30 + " App Starting " + "-"*30)
     # Check for SPACE_HOST and SPACE_ID at startup for information
     space_host_startup = os.getenv("SPACE_HOST")
-    space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
+    space_id_startup = "l3xv/Final_Assignment_Template"
 
     if space_host_startup:
         print(f"✅ SPACE_HOST found: {space_host_startup}")