running pre-commit hooks

src/senselab/audio/tasks/features_extraction/__init__.py

@@ -1 +1 @@
-""".. include:: ./doc.md"""  # noqa: D415
+""".. include:: ./doc.md"""  # noqa: D415

src/senselab/audio/tasks/features_extraction/api.py

@@ -1,6 +1,5 @@
 """This module provides functions for extracting features from audio files."""
 
-
 from typing import Any, Dict, List
 
 import pydra
@@ -24,37 +23,42 @@ def extract_features_from_audios(audios: List[Audio], plugin: str = "cf") -> Lis
         List[Dict[str, Any]]: The list of feature dictionaries for each audio.
     """
     # opensmile
-    extract_opensmile_features_from_audios_pt = pydra.mark.task(
-        extract_opensmile_features_from_audios)
+    extract_opensmile_features_from_audios_pt = pydra.mark.task(extract_opensmile_features_from_audios)
     # praat_parselmouth
-    extract_praat_parselmouth_features_from_audios_pt = pydra.mark.task(
-        extract_praat_parselmouth_features_from_audios)
+    extract_praat_parselmouth_features_from_audios_pt = pydra.mark.task(extract_praat_parselmouth_features_from_audios)
     # torchaudio
     extract_torchaudio_features_from_audios_pt = pydra.mark.task(extract_torchaudio_features_from_audios)
     # torchaudio_squim
-    extract_objective_quality_features_from_audios_pt = pydra.mark.task(
-        extract_objective_quality_features_from_audios)
+    extract_objective_quality_features_from_audios_pt = pydra.mark.task(extract_objective_quality_features_from_audios)
 
     formatted_audios = [[audio] for audio in audios]
 
     wf = pydra.Workflow(name="wf", input_spec=["x"])
     wf.split("x", x=formatted_audios)
-    wf.add(extract_opensmile_features_from_audios_pt(name="extract_opensmile_features_from_audios_pt", 
-                                                     audios=wf.lzin.x))
-    wf.add(extract_praat_parselmouth_features_from_audios_pt(name="extract_praat_parselmouth_features_from_audios_pt", 
-                                                             audios=wf.lzin.x))
-    wf.add(extract_torchaudio_features_from_audios_pt(name="extract_torchaudio_features_from_audios_pt", 
-                                                       audios=wf.lzin.x))
-    wf.add(extract_objective_quality_features_from_audios_pt(name="extract_objective_quality_features_from_audios_pt", 
-                                                             audio_list=wf.lzin.x))
+    wf.add(
+        extract_opensmile_features_from_audios_pt(name="extract_opensmile_features_from_audios_pt", audios=wf.lzin.x)
+    )
+    wf.add(
+        extract_praat_parselmouth_features_from_audios_pt(
+            name="extract_praat_parselmouth_features_from_audios_pt", audios=wf.lzin.x
+        )
+    )
+    wf.add(
+        extract_torchaudio_features_from_audios_pt(name="extract_torchaudio_features_from_audios_pt", audios=wf.lzin.x)
+    )
+    wf.add(
+        extract_objective_quality_features_from_audios_pt(
+            name="extract_objective_quality_features_from_audios_pt", audio_list=wf.lzin.x
+        )
+    )
 
     # setting multiple workflow outputs
     wf.set_output(
         [
-            ("opensmile_out", wf.extract_opensmile_features_from_audios_pt.lzout.out), 
-            ("praat_parselmouth_out", wf.extract_praat_parselmouth_features_from_audios_pt.lzout.out), 
+            ("opensmile_out", wf.extract_opensmile_features_from_audios_pt.lzout.out),
+            ("praat_parselmouth_out", wf.extract_praat_parselmouth_features_from_audios_pt.lzout.out),
             ("torchaudio_out", wf.extract_torchaudio_features_from_audios_pt.lzout.out),
-            ("torchaudio_squim_out", wf.extract_objective_quality_features_from_audios_pt.lzout.out)
+            ("torchaudio_squim_out", wf.extract_objective_quality_features_from_audios_pt.lzout.out),
         ]
     )
 
@@ -67,17 +71,15 @@ def extract_features_from_audios(audios: List[Audio], plugin: str = "cf") -> Lis
     for output in outputs:
         formatted_output_item = {
             "opensmile": output.output.opensmile_out[0],
-            "praat_parselmouth": output.output.praat_parselmouth_out['praat_parselmouth'][0],
-            "torchaudio": output.output.torchaudio_out[0]['torchaudio'],
+            "praat_parselmouth": output.output.praat_parselmouth_out["praat_parselmouth"][0],
+            "torchaudio": output.output.torchaudio_out[0]["torchaudio"],
             "torchaudio_squim": {
-                "stoi": output.output.torchaudio_squim_out['stoi'][0],
-                "pesq": output.output.torchaudio_squim_out['pesq'][0],
-                "si_sdr": output.output.torchaudio_squim_out['si_sdr'][0]
-            }
+                "stoi": output.output.torchaudio_squim_out["stoi"][0],
+                "pesq": output.output.torchaudio_squim_out["pesq"][0],
+                "si_sdr": output.output.torchaudio_squim_out["si_sdr"][0],
+            },
         }
 
         formatted_output.append(formatted_output_item)
 
     return formatted_output
-
-

src/senselab/audio/tasks/features_extraction/doc.md

@@ -10,6 +10,6 @@ This module provides the API of the senselab audio features extraction.
 
 Features can be extracted using `opensmile`, `praat-parselmouth`, `torchaudio`, and `torchaudio-squim`.
 We are working to facilitate the way to extract features in a meaningful way.
-Also, we are working to optimize these utilities. 
+Also, we are working to optimize these utilities.
 
 **STAY TUNED**.

src/senselab/audio/tasks/features_extraction/praat_parselmouth.py

@@ -432,12 +432,12 @@ def extract_pitch_descriptors(
         if current_frame is not None:
             current_function_name = current_frame.f_code.co_name
             logger.error(f'Error in "{current_function_name}": \n' + str(e))
-        return {f"mean_f0_{unit.lower()}": float("nan"), 
-                f"stdev_f0_{unit.lower()}": float("nan")}
+        return {f"mean_f0_{unit.lower()}": float("nan"), f"stdev_f0_{unit.lower()}": float("nan")}
 
 
-def extract_intensity_descriptors(snd: Union[parselmouth.Sound, Path, Audio], 
-                                  floor: float, frame_shift: float) -> Dict[str, float]:
+def extract_intensity_descriptors(
+    snd: Union[parselmouth.Sound, Path, Audio], floor: float, frame_shift: float
+) -> Dict[str, float]:
     """Extract Intensity Features.
 
     Function to extract key intensity information from a given sound object.
@@ -543,7 +543,7 @@ def extract_harmonicity_descriptors(
         if current_frame is not None:
             current_function_name = current_frame.f_code.co_name
             logger.error(f'Error in "{current_function_name}": \n' + str(e))
-        
+
         return {"hnr_db_mean": float("nan"), "hnr_db_std_dev": float("nan")}
 
 
@@ -924,10 +924,12 @@ def extract_spectral_moments(
         if current_frame is not None:
             current_function_name = current_frame.f_code.co_name
             logger.error(f'Error in "{current_function_name}": \n' + str(e))
-        return {"spectral_gravity": np.nan, 
-                "spectral_std_dev": np.nan, 
-                "spectral_skewness": np.nan, 
-                "spectral_kurtosis": np.nan}
+        return {
+            "spectral_gravity": np.nan,
+            "spectral_std_dev": np.nan,
+            "spectral_skewness": np.nan,
+            "spectral_kurtosis": np.nan,
+        }
 
 
 ### More functions ###
@@ -1044,9 +1046,9 @@ def _extract_shimmer(type: str, sound: parselmouth.Sound, point_process: parselm
 
 
 ### Wrapper ###
-def extract_praat_parselmouth_features_from_audios(audios: List[Audio], 
-                                 cache_dir: Optional[str] = None,
-                                 plugin: str = "cf") -> dict:
+def extract_praat_parselmouth_features_from_audios(
+    audios: List[Audio], cache_dir: Optional[str] = None, plugin: str = "cf"
+) -> dict:
     """Extract features from a list of Audio objects and return a JSON-like dictionary.
 
     Args:
@@ -1073,10 +1075,12 @@ def extract_praat_parselmouth_features_from_audios(audios: List[Audio],
 
     def _extract_pitch_floor(pitch_values_out: dict) -> float:
         return pitch_values_out["pitch_floor"]
+
     _extract_pitch_floor_pt = pydra.mark.task(_extract_pitch_floor)
 
     def _extract_pitch_ceiling(pitch_values_out: dict) -> float:
         return pitch_values_out["pitch_ceiling"]
+
     _extract_pitch_ceiling_pt = pydra.mark.task(_extract_pitch_ceiling)
 
     # Create the workflow
@@ -1101,7 +1105,7 @@ def _extract_pitch_ceiling(pitch_values_out: dict) -> float:
             floor=wf._extract_pitch_floor_pt.lzout.out,
             ceiling=wf._extract_pitch_ceiling_pt.lzout.out,
             frame_shift=time_step,
-            unit=unit
+            unit=unit,
         )
     )
     wf.add(

src/senselab/audio/tasks/features_extraction/torchaudio.py

@@ -177,19 +177,14 @@ def extract_pitch_from_audios(
         pitches.append(
             {
                 "pitch": torchaudio.functional.detect_pitch_frequency(
-                    audio.waveform, 
-                    sample_rate=audio.sampling_rate, 
-                    freq_low=freq_low, 
-                    freq_high=freq_high
+                    audio.waveform, sample_rate=audio.sampling_rate, freq_low=freq_low, freq_high=freq_high
                 ).squeeze(0)
             }
         )
     return pitches
 
 
-
-def extract_torchaudio_features_from_audios(audios: List[Audio], 
-                                            plugin: str = "cf") -> List[Dict[str, Any]]:
+def extract_torchaudio_features_from_audios(audios: List[Audio], plugin: str = "cf") -> List[Dict[str, Any]]:
     """Extract torchaudio features from a list of audio objects.
 
     Args:
@@ -200,26 +195,19 @@ def extract_torchaudio_features_from_audios(audios: List[Audio],
         List[Dict[str, Any]]: The list of feature dictionaries for each audio.
     """
     extract_pitch_from_audios_pt = pydra.mark.task(extract_pitch_from_audios)
-    extract_mel_filter_bank_from_audios_pt = pydra.mark.task(
-        extract_mel_filter_bank_from_audios)
+    extract_mel_filter_bank_from_audios_pt = pydra.mark.task(extract_mel_filter_bank_from_audios)
     extract_mfcc_from_audios_pt = pydra.mark.task(extract_mfcc_from_audios)
-    extract_mel_spectrogram_from_audios_pt = pydra.mark.task(
-        extract_mel_spectrogram_from_audios)
+    extract_mel_spectrogram_from_audios_pt = pydra.mark.task(extract_mel_spectrogram_from_audios)
     extract_spectrogram_from_audios_pt = pydra.mark.task(extract_spectrogram_from_audios)
 
     formatted_audios = [[audio] for audio in audios]
     wf = pydra.Workflow(name="wf", input_spec=["x"])
     wf.split("x", x=formatted_audios)
-    wf.add(extract_pitch_from_audios_pt(name="extract_pitch_from_audios_pt", 
-                                                     audios=wf.lzin.x))
-    wf.add(extract_mel_filter_bank_from_audios_pt(name="extract_mel_filter_bank_from_audios_pt", 
-                                                     audios=wf.lzin.x))
-    wf.add(extract_mfcc_from_audios_pt(name="extract_mfcc_from_audios_pt", 
-                                                     audios=wf.lzin.x))
-    wf.add(extract_mel_spectrogram_from_audios_pt(name="extract_mel_spectrogram_from_audios_pt", 
-                                                     audios=wf.lzin.x))
-    wf.add(extract_spectrogram_from_audios_pt(name="extract_spectrogram_from_audios_pt", 
-                                                     audios=wf.lzin.x))
+    wf.add(extract_pitch_from_audios_pt(name="extract_pitch_from_audios_pt", audios=wf.lzin.x))
+    wf.add(extract_mel_filter_bank_from_audios_pt(name="extract_mel_filter_bank_from_audios_pt", audios=wf.lzin.x))
+    wf.add(extract_mfcc_from_audios_pt(name="extract_mfcc_from_audios_pt", audios=wf.lzin.x))
+    wf.add(extract_mel_spectrogram_from_audios_pt(name="extract_mel_spectrogram_from_audios_pt", audios=wf.lzin.x))
+    wf.add(extract_spectrogram_from_audios_pt(name="extract_spectrogram_from_audios_pt", audios=wf.lzin.x))
 
     # setting multiple workflow outputs
     wf.set_output(
@@ -241,11 +229,11 @@ def extract_torchaudio_features_from_audios(audios: List[Audio],
     for output in outputs:
         formatted_output_item = {
             "torchaudio": {
-                "pitch": output.output.pitch_out[0]['pitch'],
-                "mel_filter_bank": output.output.mel_filter_bank_out[0]['mel_filter_bank'],
-                "mfcc": output.output.mfcc_out[0]['mfcc'],
-                "mel_spectrogram": output.output.mel_spectrogram_out[0]['mel_spectrogram'],
-                "spectrogram": output.output.spectrogram_out[0]['spectrogram'],
+                "pitch": output.output.pitch_out[0]["pitch"],
+                "mel_filter_bank": output.output.mel_filter_bank_out[0]["mel_filter_bank"],
+                "mfcc": output.output.mfcc_out[0]["mfcc"],
+                "mel_spectrogram": output.output.mel_spectrogram_out[0]["mel_spectrogram"],
+                "spectrogram": output.output.spectrogram_out[0]["spectrogram"],
             }
         }
 

src/senselab/audio/tasks/features_extraction/torchaudio_squim.py

@@ -9,6 +9,7 @@
 objective_model = SQUIM_OBJECTIVE.get_model()
 subjective_model = SQUIM_SUBJECTIVE.get_model()
 
+
 def extract_objective_quality_features_from_audios(audio_list: List[Audio]) -> Dict[str, Any]:
     """Extracts objective audio features from a list of Audio objects.
 

src/tests/audio/tasks/features_extraction_test.py

@@ -122,10 +122,9 @@ def test_extract_spectral_moments(resampled_mono_audio_sample: Audio) -> None:
         resampled_mono_audio_sample, floor=75.0, ceiling=500.0, window_size=0.025, frame_shift=0.01
     )
     assert isinstance(result, dict)
-    assert all(key in result for key in ["spectral_gravity", 
-                                         "spectral_std_dev", 
-                                         "spectral_skewness", 
-                                         "spectral_kurtosis"])
+    assert all(
+        key in result for key in ["spectral_gravity", "spectral_std_dev", "spectral_skewness", "spectral_kurtosis"]
+    )
     assert all(isinstance(result[key], float) for key in result)
 
 

tutorials/audio/00_getting_started.ipynb

@@ -66,7 +66,7 @@
             "metadata": {},
             "source": [
                 "## Downmixing audio clips to mono\n",
-                "Want to downmix your audio to mono? It has neve been that **EASY!**! Here’s how:"
+                "Want to downmix your audio to mono? It has neve been that **EASY!**! Here\u2019s how:"
             ]
         },
         {
@@ -206,7 +206,7 @@
             "metadata": {},
             "source": [
                 "## Speaker Embeddings\n",
-                "Need to get unique speaker signatures? **EASY!** Here’s how:"
+                "Need to get unique speaker signatures? **EASY!** Here\u2019s how:"
             ]
         },
         {
@@ -251,7 +251,7 @@
             "metadata": {},
             "source": [
                 "## Audio Augmentation\n",
-                "Need to augment your audio data? **EASY!**! Here’s how:"
+                "Need to augment your audio data? **EASY!**! Here\u2019s how:"
             ]
         },
         {
@@ -296,7 +296,7 @@
             "metadata": {},
             "source": [
                 "## Speech Enhancement\n",
-                "Need to clean up your audio? **EASY!** Here’s how:"
+                "Need to clean up your audio? **EASY!** Here\u2019s how:"
             ]
         },
         {