Add Google STT

theagent.py

@@ -11,10 +11,11 @@ from smolagents import (
     PythonInterpreterTool,
     VisitWebpageTool,
     OpenAIServerModel,
-    SpeechToTextTool,
     Tool,
 )
 
+from tools import EnglishSpeechToTextTool
+
 
 cache = {
     "How many studio albums were published by Mercedes Sosa between 2000 and 2009 (included)? You can use the latest 2022 version of english wikipedia.": "4",  # wrong, should be 3
@@ -26,7 +27,7 @@ cache = {
     "Examine the video at https://www.youtube.com/watch?v=1htKBjuUWec.": "Extremely",  # correct
     "What is the surname of the equine veterinarian mentioned in 1.E Exercises from the chemistry materials licensed by Marisa Alviar-Agnew & Henry Agnew under the CK-12 license in LibreText's Introductory Chemistry materials as compiled 08/21/2023?": "Louvrier",  # correct
     "I'm making a grocery list for my mom, but she's a professor of botany and she's a real stickler when it comes to categorizing things. I need to add different foods to different categories on the grocery list, but if I make a mistake, she won't buy anything inserted in the wrong category. Here's the list I have so far:": "wrong",  # wrong
-    "Hi, I'm making a pie but I could use some help with my shopping list. I have everything I need for the crust, but I'm not sure about the filling.": "cornstarch, lemon juice, salt, strawberries, sugar",  # wrong
+    "Hi, I'm making a pie but I could use some help with my shopping list. I have everything I need for the crust, but I'm not sure about the filling.": None,  # wrong
     "Who did the actor who played Ray in the Polish-language version of Everybody Loves Raymond play in Magda M.? Give only the first name.": "Wojciech",  # correct
     "What is the final numeric output from the attached Python code?": "0",  # correct
     "How many at bats did the Yankee with the most walks in the 1977 regular season have that same season?": "519",  # correct
@@ -35,7 +36,7 @@ cache = {
     "Where were the Vietnamese specimens described by Kuznetzov in Nedoshivina's 2010 paper eventually deposited? Just give me the city name without abbreviations.": "Saint Petersburg",  # correct, but not always
     "What country had the least number of athletes at the 1928 Summer Olympics? If there's a tie for a number of athletes, return the first in alphabetical order. Give the IOC country code as your answer.": "CUB",  # correct
     "Who are the pitchers with the number before and after Taishō Tamai's number as of July 2023? Give them to me in the form Pitcher Before, Pitcher After, use their last names only, in Roman characters.": "wrong",  # wrong
-    "The attached Excel file contains the sales of menu items for a local fast-food chain. What were the total sales that the chain made from food (not including drinks)? Express your answer in USD with two decimal places.": "wrong",  # wrong
+    "The attached Excel file contains the sales of menu items for a local fast-food chain. What were the total sales that the chain made from food (not including drinks)? Express your answer in USD with two decimal places.": None,  # wrong
     "What is the first name of the only Malko Competition recipient from the 20th Century (after 1977) whose nationality on record is a country that no longer exists?": "Claus",  # correct
 }
 
@@ -82,14 +83,22 @@ class BasicAgent:
 
         # Gemini
         model = OpenAIServerModel(
-            model_id="gemini-2.5-flash-preview-04-17",
-            # model_id="gemini-2.0-flash",
+            # model_id="gemini-2.5-flash-preview-04-17",
+            model_id="gemini-2.0-flash",
             api_base="https://generativelanguage.googleapis.com/v1beta/openai/",
             api_key=os.getenv("GEMINI_API_TOKEN"),
             # reasoning_effort="low",  # Only for 2.5
             # temperature=0,
         )
 
+        model25 = OpenAIServerModel(
+            model_id="gemini-2.5-flash-preview-04-17",
+            api_base="https://generativelanguage.googleapis.com/v1beta/openai/",
+            api_key=os.getenv("GEMINI_API_TOKEN"),
+            reasoning_effort="low",  # Only for 2.5
+            # temperature=0,
+        )
+
         web_agent = ToolCallingAgent(
             tools=[DuckDuckGoSearchTool(), VisitWebpageTool()],
             model=model,
@@ -98,20 +107,34 @@ class BasicAgent:
             description="Runs web searches for you. Give it your query as an argument.",
         )
 
+        chess_analyzer_agent = ToolCallingAgent(
+            tools=[],
+            model=model25,
+            max_steps=3,
+            name="analyze_chess_position",
+            description="Analyze a given chess position for best one move to checkmate. Give it the board (included in `task` argument) in algebraic notation",
+        )
+        chess_analyzer_agent.prompt_templates["system_prompt"] = (
+            "You are an expert chess puzzle solver. You will be given a chess puzzle in algebraic notation, that you have to provide the step to checkmate. You also have to answer in algebraic notation."
+        )
+
         excel_summarizer_agent = CodeAgent(
             tools=[],
             model=model,
             max_steps=3,
             name="summarize_excel_sheet",
             additional_authorized_imports=["numpy", "pandas"],
-            description="Summarize an sheet using pandas.read_excel(). Give it the path to the excel sheet as an argument",
+            description="Summarize an sheet. Give it the path to the excel sheet as an argument to `task`",
+        )
+        excel_summarizer_agent.prompt_templates["system_prompt"] = (
+            "You are an expert excel file analyzer using. You read the file using pandas.read_excel(). You answer by giving descriptive statistics of excel file."
         )
 
         manager_agent = CodeAgent(
-            tools=[SpeechToTextTool(), ExecutePythonTool()],
-            #tools=[ExecutePythonTool()],
+            tools=[EnglishSpeechToTextTool(), ExecutePythonTool()],
+            # tools=[ExecutePythonTool()],
             model=model,
-            managed_agents=[web_agent, excel_summarizer_agent],
+            managed_agents=[web_agent, excel_summarizer_agent, chess_analyzer_agent],
             additional_authorized_imports=["time", "numpy", "pandas", "os"],
         )
         self.agent = manager_agent

tools.py

@@ -0,0 +1,57 @@
+from smolagents import SpeechToTextTool, Tool
+
+class EnglishSpeechToTextTool(SpeechToTextTool):
+    def encode(self, audio):
+        from smolagents.agent_types import AgentAudio
+
+        audio = AgentAudio(audio).to_raw()
+        return self.pre_processor(audio, return_tensors="pt", sampling_rate=16_000)
+
+    def forward(self, inputs):
+        return self.model.generate(inputs["input_features"], language="en")
+
+    def decode(self, outputs):
+        return "## Transcription\n\n" + self.pre_processor.batch_decode(outputs, skip_special_tokens=True)[0]
+
+
+class GoogleSTTTool(Tool):
+    description = "This is a tool that transcribes an audio into text. It returns the transcribed text."
+    name = "transcriber"
+    inputs = {
+        "audio": {
+            "type": "audio",
+            "description": "The audio to transcribe. Can be a local path, an url, or a tensor.",
+        }
+    }
+    output_type = "string"
+
+    def forward(self, inputs):
+        from google.cloud.speech_v2 import SpeechClient
+        from google.cloud.speech_v2.types import cloud_speech
+        audio_file = inputs["audio"]
+
+        with open(audio_file, "rb") as f:
+            audio_content = f.read()
+
+        # Instantiates a client
+        client = SpeechClient()
+
+        config = cloud_speech.RecognitionConfig(
+            auto_decoding_config=cloud_speech.AutoDetectDecodingConfig(),
+            language_codes=["en-US"],
+            model="long",
+        )
+
+        request = cloud_speech.RecognizeRequest(
+            recognizer=f"projects/{PROJECT_ID}/locations/global/recognizers/_",
+            config=config,
+            content=audio_content,
+        )
+
+        # Transcribes the audio into text
+        response = client.recognize(request=request)
+
+        for result in response.results:
+            print(f"Transcript: {result.alternatives[0].transcript}")
+
+        return response