diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index dd05e9e..621d23a 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -41,3 +41,26 @@ repos:
             # "--check-hidden"
         ]
         # You can create a .codespellignore file with one word per line for words to ignore.
+
+
+# --- Use a "local" hook for mypy ---
+-   repo: local
+    hooks:
+      - id: mypy
+        name: mypy
+        # The command to execute. It's expected to be found in the PATH
+        # of the environment where you run `git commit`.
+        entry: mypy
+        args: ["--explicit-package-bases", "source/"]
+        # 'system' tells pre-commit to find the command in the current environment
+        # instead of building an isolated one.
+        language: system
+        # Specify which files to run on.
+        types: [python]
+        pass_filenames: false
+        # You might still need args if not configured elsewhere, but a mypy.ini is better.
+        # args: [--config-file=mypy.ini]
+        # `pass_filenames: false` can be useful here if you want mypy to analyze the
+        # whole project as configured in mypy.ini, rather than just the changed files.
+        # Try without it first.
+        # pass_filenames: false
diff --git a/mypy.ini b/mypy.ini
new file mode 100644
index 0000000..d888af5
--- /dev/null
+++ b/mypy.ini
@@ -0,0 +1,22 @@
+[mypy]
+# Basic configuration
+python_version = 3.11
+warn_return_any = True
+warn_unused_ignores = True
+mypy_path = source/
+
+# This is often needed when starting out, especially with libraries that lack stubs
+ignore_missing_imports = True
+
+# You can get stricter over time by removing ignore_missing_imports
+# and setting flags like these:
+# disallow_untyped_defs = True
+# disallow_any_unimported = True
+
+# Tell mypy to ignore errors from third-party libraries if they are not typed
+[mypy-torch.*]
+ignore_missing_imports = True
+[mypy-torchaudio.*]
+ignore_missing_imports = True
+[mypy-pytest.*]
+ignore_missing_imports = True
diff --git a/source/nodes/nodes_audio.py b/source/nodes/nodes_audio.py
index 0234e0e..94a5f48 100644
--- a/source/nodes/nodes_audio.py
+++ b/source/nodes/nodes_audio.py
@@ -7,6 +7,7 @@
 import torch
 import torchaudio
 import torchaudio.transforms as T
+from typing import Optional, Dict, Any
 from .utils.aligner import AudioBatchAligner
 from .utils.logger import main_logger
 from .utils.misc import parse_time_to_seconds, parse_note_to_frequency
@@ -429,7 +430,7 @@ class AudioBlend:
     UNIQUE_NAME = "SET_AudioBlend"
     DISPLAY_NAME = "Audio Blend"
 
-    def blend_audio(self, audio1: dict, gain1: float, gain2: float, audio2: dict = None):
+    def blend_audio(self, audio1: dict, gain1: float, gain2: float, audio2: Optional[Dict[str, Any]] = None):
         if audio2 is None:
             # Handle the simple case where audio2 is not provided
             logger.info(f"Blending audio1 only with gain {gain1}.")
@@ -609,6 +610,7 @@ class AudioTestSignalGenerator:
             waveform_multichannel = torch.zeros((batch_size, channels, num_samples))
 
         # Final clip to ensure [-1, 1] range, as some operations might exceed it slightly
+        assert waveform_multichannel is not None, "Waveform should have been created by now"
         final_waveform = torch.clamp(waveform_multichannel, -1.0, 1.0)
 
         logger.info(f"Generated '{waveform_type}' signal. Shape: {final_waveform.shape}, SR: {sample_rate}Hz")
diff --git a/source/nodes/utils/logger.py b/source/nodes/utils/logger.py
index 846f090..d63c596 100644
--- a/source/nodes/utils/logger.py
+++ b/source/nodes/utils/logger.py
@@ -2,21 +2,28 @@
 # Copyright (c) 2025 Instituto Nacional de Tecnologïa Industrial
 # License: GPL-3.0
 # Project: ComfyUI-AudioBatch
+from __future__ import annotations  # Good practice
+import logging
 import os
 import sys
-import logging
+from typing import Any
 from .misc import NODES_NAME, NODES_DEBUG_VAR
 
-no_colorama = False
+
+# 1. Initialize variables with the `Any` type.
+#    This tells mypy not to make assumptions about their specific class.
+Fore: Any
+Back: Any
+Style: Any
+
+# 2. Perform the runtime import logic as before.
 try:
     from colorama import init as colorama_init, Fore, Back, Style
-except ImportError:
-    no_colorama = True
-# If colorama isn't installed use an ANSI basic replacement
-if no_colorama:
-    from .ansi import Fore, Back, Style  # noqa: F811
-else:
     colorama_init()
+except ImportError:
+    # If colorama is not available, import our fallback.
+    # mypy will now allow this assignment because the variables were declared as Any.
+    from .ansi import Fore, Back, Style
 
 
 class CustomFormatter(logging.Formatter):
diff --git a/source/tests/test_audio_processing_nodes.py b/source/tests/test_audio_processing_nodes.py
index 7fa8285..07f797a 100644
--- a/source/tests/test_audio_processing_nodes.py
+++ b/source/tests/test_audio_processing_nodes.py
@@ -144,7 +144,7 @@ def get_peak_frequency(waveform: torch.Tensor, sample_rate: int) -> float:
     peak_index = torch.argmax(torch.abs(fft_result))
     # Get the frequency corresponding to that peak index
     peak_freq = freq_bins[peak_index].item()
-    return peak_freq
+    return float(peak_freq)
 
 
 def test_resampler_preserves_frequency_content(resampler_node):
diff --git a/source/tests/test_audio_test_signal_generator.py b/source/tests/test_audio_test_signal_generator.py
index 2be96b1..d635f6d 100644
--- a/source/tests/test_audio_test_signal_generator.py
+++ b/source/tests/test_audio_test_signal_generator.py
@@ -42,7 +42,7 @@ def get_peak_frequency(waveform: torch.Tensor, sample_rate: int) -> float:
     peak_index = torch.argmax(torch.abs(fft_result))
     # Get the frequency corresponding to that peak index
     peak_freq = freq_bins[peak_index].item()
-    return peak_freq
+    return float(peak_freq)
 
 
 # --- Test Cases ---