### Install FFmpeg on Google Colab

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

This snippet verifies the installation of FFmpeg on Google Colab. Ensure FFmpeg and FFprobe binaries are in your PATH.

```python
from analyzeAudio.ffmpeg import verifyFFmpegColab
verifyFFmpegColab()
```

--------------------------------

### Example Usage of analyzeSpectralFlatness

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseSpectrogram.md

Demonstrates how to use the analyzeSpectralFlatness function and compute the mean of the resulting spectral flatness values.

```python
from analyzeAudio.analyzersUseSpectrogram import analyzeSpectralFlatness

spectral_flatness_db = analyzeSpectralFlatness(spectrogram_magnitude)
sf_mean = spectral_flatness_db.mean()
```

--------------------------------

### Set Multiprocessing Spawn Method

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

Ensure consistent multiprocessing behavior across platforms by setting the start method to 'spawn' if it hasn't been configured already. This is handled by the audioAspectsRegistry module.

```python
from multiprocessing import set_start_method
import contextlib
with contextlib.suppress(RuntimeError):
    set_start_method('spawn')
```

--------------------------------

### List Available Audio Aspects and Contests

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Use these Python functions to retrieve lists of available audio aspects and contests from the installed package. Ensure the analyzeAudio package is installed.

```python
from analyzeAudio import getListAvailableAudioAspects, getListAvailableAudioContests

print(getListAvailableAudioAspects())
print(getListAvailableAudioContests())
```

--------------------------------

### Command-Line Interface for Audio Aspects and Contests

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

These are the corresponding terminal commands to list available audio aspects and contests. These commands are available if the package is installed and configured in your system's PATH.

```powershell
whatAspects
whatContests
```

--------------------------------

### File Format Example: Delimited Text Output

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/misfit.md

Illustrates the structure of the output file generated by dataTabularTOpathFilenameDelimited, showing header and data rows separated by a tab delimiter.

```text
pathFilename	LUFS integrated	Spectral Flatness mean
/path/to/track1.wav	-23.5	0.12
/path/to/track2.wav	-21.2	0.15
```

--------------------------------

### Python Example: Writing Tabular Data to Delimited Files

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/misfit.md

Demonstrates how to use dataTabularTOpathFilenameDelimited to write analysis results to TSV, CSV, and a file without a header. Ensure necessary imports are present.

```python
from analyzeAudio import analyzeAudioListPathFilenames, dataTabularTOpathFilenameDelimited

audio_files = ['/path/to/track1.wav', '/path/to/track2.wav']
aspects = ['LUFS integrated', 'Spectral Flatness mean']

rows = analyzeAudioListPathFilenames(audio_files, aspects)

# Write as TSV (default)
dataTabularTOpathFilenameDelimited(
    'results.tsv',
    rows,
    ['pathFilename', *aspects]
)

# Write as CSV
dataTabularTOpathFilenameDelimited(
    'results.csv',
    rows,
    ['pathFilename', *aspects],
    delimiterOutput=','
)

# No header
dataTabularTOpathFilenameDelimited(
    'results_no_header.txt',
    rows,
    tableColumns=None  # Suppresses header
)
```

--------------------------------

### Get List of Available Audio Aspects

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/aspects-registry.md

Retrieve a sorted list of all available audio analysis aspects at runtime. This is useful for understanding the full scope of metrics provided by the library.

```python
from analyzeAudio import getListAvailableAudioAspects

all_aspects = getListAvailableAudioAspects()
print(sorted(all_aspects))
```

--------------------------------

### Get Detailed Per-Frame Arrays

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Retrieve detailed, per-frame analysis results for specific audio aspects. This is useful for analyzing dynamic changes over time, as opposed to overall summary values.

```python
from analyzeAudio.analyzersUseFilename import (
    analyzeLUFSIntegratedOverall,
    analyzeLUFSMomentary,
)

integrated = analyzeLUFSIntegratedOverall("voice.wav")
momentaryFrames = analyzeLUFSMomentary("voice.wav")
```

--------------------------------

### Get Mean Signal-to-Distortion Ratio

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseFilename.md

Computes the mean signal-to-distortion ratio between two audio files. This metric is useful for evaluating audio processing effects.

```python
def getSDRmean(
    pathFilenameAlfa: str | PathLike[Any],
    pathFilenameBeta: str | PathLike[Any]
) -> float | None:
    pass
```

--------------------------------

### Initialize Analyzer Registry

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The audioAspects registry is automatically populated when analyzer sub-modules are imported. This snippet shows how to access the registry and list available aspects.

```python
from analyzeAudio import (
    audioAspects,
    getListAvailableAudioAspects,
    analyzeAudioFile
)

# Registry is now populated
available_aspects = getListAvailableAudioAspects()
```

--------------------------------

### Analyze Integrated Loudness (LUFS)

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseFilename.md

Computes the integrated programme loudness of an audio file. Use this function to get the overall loudness of an audio file in LUFS.

```python
from analyzeAudio import analyzeAudioFile

result = analyzeAudioFile('audio.wav', ['LUFS integrated'])
# Returns something like [-23.5]
```

--------------------------------

### Analyze Tensor-Based Metrics (Single File)

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Compute audio quality metrics on PyTorch tensors for a single audio file. Requires the audio tensor, sample rate, and an optional flag for CPU usage.

```python
def analyzer(tensorAudio: torch.Tensor, sampleRate: int, pytorchOnCPU: bool | None, **kwargs) -> float
```

--------------------------------

### Measure Multiple Audio Files from Path

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Analyze multiple WAV files from a specified directory. Results are returned as rows, with each row containing the filename and its measurements. Rows are returned as files finish processing.

```python
from pathlib import Path
from analyzeAudio import analyzeAudioListPathFilenames

listPathFilenames = tuple(Path("audio").glob("*.wav"))
listAspectNames = [
    "LUFS integrated",
    "LUFS loudness range",
    "true_peak maximum",
]

rows = analyzeAudioListPathFilenames(
    listPathFilenames,
    listAspectNames,
    CPUlimit=4,
)

for row in rows:
    print(row)
```

--------------------------------

### Analyze Tensor-Based Metrics (Comparison)

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Compute audio quality metrics on PyTorch tensors for comparing two or three audio files. These functions take the respective audio tensors as input.

```python
def analyzer(tensorAudioAlfa: torch.Tensor, tensorAudioBeta: torch.Tensor, **kwargs) -> float
def analyzer(tensorAudioAlfa: torch.Tensor, tensorAudioBeta: torch.Tensor, tensorAudioGamma: torch.Tensor, sampleRate: int, **kwargs) -> float
```

--------------------------------

### Accessing Audio Aspects Registry

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/audioAspectsRegistry.md

Demonstrates how to access metadata for a registered audio aspect from the 'audioAspects' registry. Shows how to retrieve the analyzer function and its parameter names, and how to call an analyzer directly if needed.

```python
from analyzeAudio import audioAspects

# Get metadata for a registered aspect
metadata = audioAspects['LUFS integrated']
analyzer_func = metadata['analyzer']
param_names = metadata['analyzerParameters']

# Call an analyzer directly if it needs more than pathFilename
# Example: SI-SDR compares two files
si_sdr_analyzer = audioAspects['SI-SDR mean']['analyzer']
result = si_sdr_analyzer(path_file1, path_file2)
```

--------------------------------

### AnalyzeAudio API Parameter Mapping

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Illustrates how `analyzeAudioFile` maps local variables to analyzer parameters. Analyzers expect parameters to be named from a fixed set of local variables available within the function.

```text
Local variables (inside analyzeAudioFile):
  - pathFilename (from argument)
  - waveform (computed from audio file)
  - sampleRate (computed from audio file)
  - spectrogram (computed from waveform)
  - spectrogramMagnitude (from spectrogram)
  - spectrogramPower (from spectrogram)
  - tensorAudio (from waveform)
  - pytorchOnCPU (computed from torch.cuda.is_available())
```

--------------------------------

### Get Mean Scale-Invariant Signal-to-Distortion Ratio

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseFilename.md

Computes the mean scale-invariant signal-to-distortion ratio between two audio files. This is useful for comparing audio with different amplitudes, as it is forgiving of scaling factors.

```python
from analyzeAudio import audioAspects

si_sdr_analyzer = audioAspects['SI-SDR mean']['analyzer']
score = si_sdr_analyzer('reference.wav', 'processed.wav')
```

--------------------------------

### analyzeTempoMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the tempo of an audio signal using Librosa, returning the mean tempo.

```APIDOC
## analyzeTempoMean analyzeTempoMean()

### Description
Analyzes the tempo of an audio signal using Librosa, returning the mean tempo.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Parameters are detailed in the source documentation, including input types and mathematical descriptions.

### Request Example
```python
# Example usage (assuming analyzeTempoMean is imported)
# mean_tempo = analyzeTempoMean()
```

### Response
#### Success Response
- Returns the mean tempo as a scalar value.

#### Response Example
```json
120.5
```
```

--------------------------------

### Get Mean Peak Signal-to-Noise Ratio

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseFilename.md

Computes the mean peak signal-to-noise ratio between two audio files. Use this to compare the quality of an original audio file against a processed version.

```python
from analyzeAudio import audioAspects

psnr_analyzer = audioAspects['Peak Signal-to-Noise Ratio mean']['analyzer']
result = psnr_analyzer('/path/to/original.wav', '/path/to/processed.wav')
```

--------------------------------

### Get List of Available Audio Aspects

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/audioAspectsRegistry.md

Retrieves a sorted list of all registered audio aspect names from the 'audioAspects' registry. This list can be used to iterate through available metrics or as input for other analysis functions.

```python
from analyzeAudio import getListAvailableAudioAspects

available = getListAvailableAudioAspects()
print(available)
# Output:
# ['Crest factor', 'Chromagram mean', 'DC offset', ...]

# Use in downstream tools
for aspect_name in available:
    print(f"Available metric: {aspect_name}")
```

--------------------------------

### Configure SRMR computation with torchmetrics parameters

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The analyzeSRMR and analyzeSRMRMean functions allow configuration of the Short-time Reverberation Modulation Energy (SRMR) metric using parameters from torchmetrics.

```python
from analyzeAudio.analyzersUseTensor import analyzeSRMRMean

# Configure SRMR computation
srmr_score = analyzeSRMRMean(
    tensorAudio,
    sampleRate=16000,
    pytorchOnCPU=True,
    n_cochlear_filters=23,
    low_freq=125.0,
    fast=True
)
```

--------------------------------

### Analyze Mel-Scaled dB Mean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Use `analyzeBleedFullMelDBMean` to get two scores, 'bleed' and 'full', representing differences in mel-scaled dB content between two spectrograms. Ensure `spectrogramMagnitudeReference` and `spectrogramMagnitudeEstimate` are defined beforehand.

```python
from analyzeAudio.contestsSpectrogram import analyzeBleedFullMelDBMean

bleedFull = analyzeBleedFullMelDBMean(
    spectrogramMagnitudeReference,
    spectrogramMagnitudeEstimate,
)
print(bleedFull.bleed, bleedFull.full)
```

--------------------------------

### Compare Audio Files Using Filename

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Use these functions to compare two audio files based on their filenames. They calculate metrics like PSNR and SDR.

```python
from analyzeAudio.analyzersUseFilename import (
    analyzePSNRmean,
    analyzeSDRmean,
    analyzeSI_SDRmean,
)

pathReference = "reference.wav"
pathEstimate = "estimate.wav"

psnr = analyzePSNRmean(pathReference, pathEstimate)
sdr = analyzeSDRmean(pathReference, pathEstimate)
si_sdr = analyzeSI_SDRmean(pathReference, pathEstimate)
```

--------------------------------

### Single File Analysis with Multiple Metrics

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Perform analysis on a single audio file for multiple specified metrics and store the results in a dictionary.

```python
from analyzeAudio import analyzeAudioFile

metrics = ['LUFS integrated', 'RMS peak', 'Spectral Flatness mean', 'SRMR mean']
results = analyzeAudioFile('myaudio.wav', metrics)
result_dict = dict(zip(metrics, results, strict=True))

print(f"Loudness: {result_dict['LUFS integrated']} LUFS")
print(f"Peak RMS: {result_dict['RMS peak']} dB")
```

--------------------------------

### Configure LogWMSE computation with frequency weighting

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The analyzeLogWMSEMean function allows customization of frequency weighting for LogWMSE calculations, including options to bypass filtering.

```python
from analyzeAudio.analyzersUseTensor import analyzeLogWMSEMean

logwmse = analyzeLogWMSEMean(
    reference,
    estimate,
    mixture,
    sampleRate=44100,
    bypass_filter=False,
    reduction='mean'
)
```

--------------------------------

### Custom Comparative Analysis

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Perform custom comparative analysis between two audio files using a specific analyzer, such as SI-SDR mean. This allows for targeted evaluation of audio processing.

```python
from analyzeAudio import audioAspects

# Get the SI-SDR analyzer
si_sdr_analyzer = audioAspects['SI-SDR mean']['analyzer']

# Compare two files
score = si_sdr_analyzer('reference.wav', 'processed.wav')
print(f"SI-SDR: {score} dB")
```

--------------------------------

### _formatTensorAudioForBatchFirstLoss

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseTensor.md

Adds a batch axis (leading singleton dimension) to audio tensors, preparing them for loss calculations that expect batch-first input. It handles 1D, 2D, and 3D inputs by prepending a batch dimension.

```APIDOC
## _formatTensorAudioForBatchFirstLoss

### Description
Adds a batch axis (leading singleton dimension) for batch-first tensor losses.

- 1D, 2D, 3D inputs: prepends one singleton batch axis
- 4D+ inputs: returned unchanged

### Parameters
#### Path Parameters
- **tensorAudio** (torch.Tensor) - Required - The input waveform tensor.

### Returns
`torch.Tensor` — The tensor with a prepended batch dimension if applicable.

### Source
`analyzeAudio/analyzersUseTensor.py` lines 188–217
```

--------------------------------

### Batch Analysis to CSV

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Analyze multiple audio files in a folder for specified metrics and save the results to a CSV file. This workflow is suitable for processing large collections of audio.

```python
from analyzeAudio import analyzeAudioListPathFilenames, dataTabularTOpathFilenameDelimited
import pathlib

audio_files = list(pathlib.Path('audio_folder').glob('*.wav'))
metrics = ['LUFS integrated', 'RMS peak']

rows = analyzeAudioListPathFilenames(audio_files, metrics, CPUlimit=4)

dataTabularTOpathFilenameDelimited(
    'output.csv',
    rows,
    ['pathFilename', *metrics],
    delimiterOutput=','
)
```

--------------------------------

### analyzeLogWMSEMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Calculates the mean Frequency-weighted Mean Squared Error (WMSE) for an audio signal using PyTorch-based metrics.

```APIDOC
## analyzeLogWMSEMean analyzeLogWMSEMean()

### Description
Calculates the mean Frequency-weighted Mean Squared Error (WMSE) for an audio signal using PyTorch-based metrics.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Tensor input requirements and full signatures with keyword-only parameters are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeLogWMSEMean is imported)
# mean_log_wmse = analyzeLogWMSEMean()
```

### Response
#### Success Response
- Returns the mean log WMSE as a scalar value.

#### Response Example
```json
0.05
```
```

--------------------------------

### analyzeAudioListPathFilenames()

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Analyzes multiple audio files in parallel. It accepts a list of file paths, a list of aspect names, and an optional CPU limit. It returns a list of lists, where each inner list contains the analysis results for a single file.

```APIDOC
## analyzeAudioListPathFilenames()

### Description
Analyzes multiple audio files in parallel. It accepts a list of file paths, a list of aspect names, and an optional CPU limit. It returns a list of lists, where each inner list contains the analysis results for a single file.

### Function Signature
`analyzeAudioListPathFilenames(listPathFilenames, listAspectNames, CPUlimit)`

### Parameters
- **listPathFilenames** (list[str]): A list of paths to the audio files to analyze.
- **listAspectNames** (list[str]): A list of names of the audio aspects to extract.
- **CPUlimit** (int, optional): The maximum number of CPUs to use for parallel processing.

### Output
- **list[list[str | float]]**: A list of lists, where each inner list contains the analysis results for a corresponding audio file.
```

--------------------------------

### Analyze Multiple Audio Files in Parallel

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyze.md

Computes requested audio aspect values for multiple audio files concurrently using process workers. Specify the number of CPU cores to use with `CPUlimit`. Results are returned in completion order, not input order.

```python
from analyzeAudio import analyzeAudioListPathFilenames, dataTabularTOpathFilenameDelimited
import pathlib

audio_files = [
    '/path/to/track1.wav',
    '/path/to/track2.wav',
    '/path/to/track3.wav',
]
aspects = ['LUFS integrated', 'Spectral Flatness mean']

rows = analyzeAudioListPathFilenames(audio_files, aspects, CPUlimit=4)

# Each row is like: ['/path/to/track1.wav', -23.5, 0.12]
# Write to CSV/TSV
dataTabularTOpathFilenameDelimited(
    'output.tsv',
    rows,
    ['pathFilename', *aspects],
    delimiterOutput='\t'
)
```

--------------------------------

### Batch Analyze Audio Files and Save to CSV

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/README.md

Analyzes multiple audio files in a directory for specified metrics and saves the results to a CSV file. Supports parallel processing up to a specified CPU limit.

```python
from analyzeAudio import analyzeAudioListPathFilenames, dataTabularTOpathFilenameDelimited
import pathlib

files = list(pathlib.Path('audio').glob('*.wav'))
metrics = ['LUFS integrated', 'Spectral Flatness mean']

rows = analyzeAudioListPathFilenames(files, metrics, CPUlimit=4)

dataTabularTOpathFilenameDelimited('output.csv', rows, ['file', *metrics], ',')
```

--------------------------------

### Compare Tensor Spectrograms

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Compare two PyTorch magnitude spectrogram tensors using spectral convergence and STFT magnitude loss functions.

```python
from analyzeAudio.contestsTensorSpectrogram import (
    analyzeSpectralConvergenceLoss,
    analyzeSTFTMagnitudeLoss,
)

spectralConvergence = analyzeSpectralConvergenceLoss(
    tensorSpectrogramMagnitudeReference,
    tensorSpectrogramMagnitudeEstimate,
)
stftMagnitude = analyzeSTFTMagnitudeLoss(
    tensorSpectrogramMagnitudeReference,
    tensorSpectrogramMagnitudeEstimate,
)
```

--------------------------------

### analyzeSpectralContrastMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the spectral contrast of an audio signal using Librosa, returning the mean spectral contrast.

```APIDOC
## analyzeSpectralContrastMean analyzeSpectralContrastMean()

### Description
Analyzes the spectral contrast of an audio signal using Librosa, returning the mean spectral contrast.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Spectrogram input types and full signatures are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeSpectralContrastMean is imported)
# mean_spectral_contrast = analyzeSpectralContrastMean()
```

### Response
#### Success Response
- Returns the mean spectral contrast as a numerical array.

#### Response Example
```json
[1.3, 1.4]
```
```

--------------------------------

### Analyze Multiple Audio Files in Parallel

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Use `analyzeAudioListPathFilenames` to analyze multiple audio files concurrently. Specify the list of file paths, aspect names, and an optional CPU limit. Returns a list of lists, where each inner list contains the results for a file.

```python
def analyzeAudioListPathFilenames(listPathFilenames: list[str], listAspectNames: list[str], CPUlimit: int | None) -> list[list[str | float]]
```

--------------------------------

### Configure L1SNR computation

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The analyzeL1SNRMean function supports parameters from the torch_l1_snr library for calculating the L1 Signal-to-Noise Ratio.

```python
from analyzeAudio.analyzersUseTensor import analyzeL1SNRMean

l1snr = analyzeL1SNRMean(
    reference_tensor,
    estimate_tensor,
    weight=1.0,
    eps=1e-3,
    l1_weight=0.0
)
```

--------------------------------

### analyzeAudioFile()

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Analyzes a single audio file to extract specified aspects. It takes the file path and a list of aspect names as input and returns a list of results, which can be strings or floats.

```APIDOC
## analyzeAudioFile()

### Description
Analyzes a single audio file to extract specified aspects. It takes the file path and a list of aspect names as input and returns a list of results, which can be strings or floats.

### Function Signature
`analyzeAudioFile(pathFilename, listAspectNames)`

### Parameters
- **pathFilename** (str): The path to the audio file to analyze.
- **listAspectNames** (list[str]): A list of names of the audio aspects to extract.

### Output
- **list[str | float]**: A list containing the extracted audio aspects. The elements can be strings or floating-point numbers.
```

--------------------------------

### Analyze Tempo from Waveform

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseWaveform.md

Estimates the tempo in beats per minute from waveform periodicity. Requires waveform data and its sample rate. Additional parameters can be passed to librosa.feature.tempo.

```python
import numpy
import soundfile
from analyzeAudio.analyzersUseWaveform import analyzeTempo

with soundfile.SoundFile('audio.wav') as f:
    sampleRate = f.samplerate
    waveform = f.read(dtype='float32').astype(numpy.float32).T

tempo = analyzeTempo(waveform, sampleRate)
```

--------------------------------

### analyzeSpectralContrast

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the spectral contrast of an audio signal using Librosa, returning the full spectral contrast matrix.

```APIDOC
## analyzeSpectralContrast analyzeSpectralContrast()

### Description
Analyzes the spectral contrast of an audio signal using Librosa, returning the full spectral contrast matrix.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Spectrogram input types and full signatures are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeSpectralContrast is imported)
# spectral_contrast_data = analyzeSpectralContrast()
```

### Response
#### Success Response
- Returns the spectral contrast as a 2D numerical array.

#### Response Example
```json
[[1.2, 1.5], [1.1, 1.3], [1.4, 1.6]]
```
```

--------------------------------

### Analyze LogWMSE Mean with PyTorch Tensors

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseTensor.md

Scores an estimated audio tensor against a reference tensor using frequency-weighted mean squared error in log scale. Requires reference, estimated, and mixture audio tensors, along with the sample rate. Optional parameters allow for custom frequency weighting and reduction modes.

```python
def analyzeLogWMSEMean(
    tensorAudioAlfa: torch.Tensor,
    tensorAudioBeta: torch.Tensor,
    tensorAudioGamma: torch.Tensor,
    sampleRate: int,
    **keywordArguments: Any
) -> float:
    # ... implementation details ...
```

--------------------------------

### Access Available Audio Metrics Registry

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/README.md

Retrieves a list of all available audio metrics registered within the analyzeAudio package. Useful for discovering supported metrics.

```python
from analyzeAudio import getListAvailableAudioAspects

all_aspects = getListAvailableAudioAspects()
print(f"Available metrics: {len(all_aspects)}")
```

--------------------------------

### analyzeTempo

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the tempo of an audio signal using Librosa, returning the full tempo time-series.

```APIDOC
## analyzeTempo analyzeTempo()

### Description
Analyzes the tempo of an audio signal using Librosa, returning the full tempo time-series.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Parameters are detailed in the source documentation, including input types and mathematical descriptions.

### Request Example
```python
# Example usage (assuming analyzeTempo is imported)
# tempo_series = analyzeTempo()
```

### Response
#### Success Response
- Returns the tempo time-series as a numerical array.

#### Response Example
```json
[120.0, 121.5, 119.0, 122.0]
```
```

--------------------------------

### Analyzer Function (File Comparison Metrics)

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Compares two audio files using FFmpeg filters to calculate metrics like Peak SNR, SDR, and SI-SDR. It takes the paths of the two audio files as input and returns a float value representing the comparison metric, or None if analysis fails.

```APIDOC
## Analyzer (File Comparison Metrics)

### Description
Compares two audio files using FFmpeg filters to calculate metrics like Peak SNR, SDR, and SI-SDR. It takes the paths of the two audio files as input and returns a float value representing the comparison metric, or None if analysis fails.

### Module
analyzeAudio.analyzersUseFilename

### Function Signature
`analyzer(pathFilenameAlfa: str | PathLike[Any], pathFilenameBeta: str | PathLike[Any]) -> float | None`

### Parameters
- **pathFilenameAlfa** (str | PathLike[Any]): The path to the first audio file.
- **pathFilenameBeta** (str | PathLike[Any]): The path to the second audio file.

### Output
- **float | None**: The calculated comparison metric, or None if the analysis could not be performed.
```

--------------------------------

### Pass librosa parameters to Waveform Analyzers

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

When using waveform analyzers like analyzeRMS, you can pass specific parameters supported by the underlying librosa implementation.

```python
from analyzeAudio.analyzersUseWaveform import analyzeRMS

# Pass librosa parameters
rms_db = analyzeRMS(waveform, frame_length=2048, hop_length=512)
```

--------------------------------

### Measure Single Audio File

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Analyze a single audio file for a list of specified aspects. Returns a dictionary of measurements. Unavailable aspects will be marked as 'not found'.

```python
from analyzeAudio import analyzeAudioFile

listAspectNames = [
    "LUFS integrated",
    "true_peak maximum",
    "RMS_level overall",
    "Spectral Flatness mean",
    "Zero Crossing Rate mean",
]

listValues = analyzeAudioFile("voice.wav", listAspectNames)
measurements = dict(zip(listAspectNames, listValues, strict=True))
print(measurements)
```

--------------------------------

### Compare Tensor Waveforms

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Compare two PyTorch waveform tensors using various signal-to-noise and loss functions. Ensure tensors are correctly imported.

```python
from analyzeAudio.contestsTensor import (
    analyzeL1SNRDBMean,
    analyzeMultiResolutionSTFTLoss,
)

l1snrdb = analyzeL1SNRDBMean(tensorReference, tensorEstimate)
mrstft = analyzeMultiResolutionSTFTLoss(tensorReference, tensorEstimate)
```

```python
from analyzeAudio.contestsTensor import analyzeLogWMSEMean

logwmse = analyzeLogWMSEMean(
    tensorReference,
    tensorEstimate,
    tensorMixture,
    sampleRate,
)
```

--------------------------------

### Configure L1SNRDB computation with regularization

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The analyzeL1SNRDBMean function enables level-matching regularization for L1SNRDB calculations, with customizable parameters.

```python
from analyzeAudio.analyzersUseTensor import analyzeL1SNRDBMean

l1snrdb = analyzeL1SNRDBMean(
    reference_tensor,
    estimate_tensor,
    lambda0=0.1,
    delta_lambda=0.9,
    use_regularization=True,
    lmin=-60.0
)
```

--------------------------------

### Compute Spectral Contrast from Spectrogram

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseSpectrogram.md

Computes octave-band peak-to-valley spectral contrast. Requires a magnitude-domain spectrogram. Additional parameters can be passed to librosa.feature.spectral_contrast.

```python
def analyzeSpectralContrast(
    spectrogramMagnitude: SpectrogramMagnitude,
    **keywordArguments: Any
) -> libturd:
    """Computes octave-band peak-to-valley contrast.

    ### Parameters

    | Parameter | Type | Required | Description |
    |-----------|------|----------|-------------|
    | spectrogramMagnitude | `SpectrogramMagnitude` | Yes | Magnitude-domain spectrogram array. Type: `numpy.ndarray[tuple[int, int, int], numpy.dtype[numpy.floating[Any]]]` |
    | keywordArguments | `Any` | No | Additional parameters forwarded to `librosa.feature.spectral_contrast()`. |

    ### Returns

    `libturd` — Framewise contrast values for each octave band.

    ### Registered Aspect

    `'Spectral Contrast mean'` — Mean of the spectral contrast trajectory.
    """
    pass
```

--------------------------------

### analyzeSpectralFlatnessMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the spectral flatness of an audio signal using Librosa, returning the mean spectral flatness.

```APIDOC
## analyzeSpectralFlatnessMean analyzeSpectralFlatnessMean()

### Description
Analyzes the spectral flatness of an audio signal using Librosa, returning the mean spectral flatness.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Spectrogram input types and full signatures are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeSpectralFlatnessMean is imported)
# mean_spectral_flatness = analyzeSpectralFlatnessMean()
```

### Response
#### Success Response
- Returns the mean spectral flatness as a scalar value.

#### Response Example
```json
0.105
```
```

--------------------------------

### analyzeTempogramMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the tempo of an audio signal using Librosa, returning the mean of the tempogram.

```APIDOC
## analyzeTempogramMean analyzeTempogramMean()

### Description
Analyzes the tempo of an audio signal using Librosa, returning the mean of the tempogram.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Parameters are detailed in the source documentation, including input types and mathematical descriptions.

### Request Example
```python
# Example usage (assuming analyzeTempogramMean is imported)
# mean_tempogram = analyzeTempogramMean()
```

### Response
#### Success Response
- Returns the mean tempo as a scalar value.

#### Response Example
```json
110.5
```
```

--------------------------------

### analyzeSpectralBandwidthMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the spectral bandwidth of an audio signal using Librosa, returning the mean spectral bandwidth.

```APIDOC
## analyzeSpectralBandwidthMean analyzeSpectralBandwidthMean()

### Description
Analyzes the spectral bandwidth of an audio signal using Librosa, returning the mean spectral bandwidth.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Spectrogram input types and full signatures are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeSpectralBandwidthMean is imported)
# mean_spectral_bandwidth = analyzeSpectralBandwidthMean()
```

### Response
#### Success Response
- Returns the mean spectral bandwidth as a scalar value.

#### Response Example
```json
1500.0
```
```

--------------------------------

### Analyzer Function (File-Based Scalar Metrics)

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Analyzes an audio file using FFprobe to extract scalar measurements like Zero Crossing, DC offset, Loudness, and Spectral Stats. It takes the file path as input and returns a single float value or None if analysis fails.

```APIDOC
## Analyzer (File-Based Scalar Metrics)

### Description
Analyzes an audio file using FFprobe to extract scalar measurements like Zero Crossing, DC offset, Loudness, and Spectral Stats. It takes the file path as input and returns a single float value or None if analysis fails.

### Module
analyzeAudio.analyzersUseFilename

### Function Signature
`analyzer(pathFilename: str | PathLike[Any]) -> float | None`

### Parameters
- **pathFilename** (str | PathLike[Any]): The path to the audio file.

### Output
- **float | None**: The scalar measurement from the audio file, or None if the analysis could not be performed.
```

--------------------------------

### analyzeRMSMean

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the Root Mean Square (RMS) energy of an audio signal using Librosa, returning the mean RMS value.

```APIDOC
## analyzeRMSMean analyzeRMSMean()

### Description
Analyzes the Root Mean Square (RMS) energy of an audio signal using Librosa, returning the mean RMS value.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Parameters are detailed in the source documentation, including input types and mathematical descriptions.

### Request Example
```python
# Example usage (assuming analyzeRMSMean is imported)
# mean_rms = analyzeRMSMean()
```

### Response
#### Success Response
- Returns the mean RMS value as a scalar.

#### Response Example
```json
0.15
```
```

--------------------------------

### Register Custom Analyzer

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

Register a new custom audio analyzer using the @registrationAudioAspect decorator. This allows the custom metric to be accessed through the audioAspects registry.

```python
from analyzeAudio.audioAspectsRegistry import registrationAudioAspect
from os import PathLike
from typing import Any

@registrationAudioAspect('My Custom Metric')
def myCustomAnalyzer(pathFilename: str | PathLike[Any]) -> float:
    # Compute metric
    return 42.0

# Now accessible via
from analyzeAudio import audioAspects
metric_value = audioAspects['My Custom Metric']['analyzer']('audio.wav')
```

--------------------------------

### ffprobeShotgunAndCache

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Internal helper function for running FFprobe with a shotgun approach and caching results. This is typically used by FFprobe-based analyzers.

```APIDOC
## ffprobeShotgunAndCache ffprobeShotgunAndCache()

### Description
Internal helper function for running FFprobe with a shotgun approach and caching results. This is typically used by FFprobe-based analyzers.

### Method
Not applicable (Internal helper function)

### Endpoint
Not applicable (Internal helper function)

### Parameters
Not explicitly documented for external use.

### Request Example
Not applicable for external use.

### Response
Not explicitly documented for external use.

### Response Example
Not applicable for external use.
```

--------------------------------

### Analyze Single Audio File

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/README.md

Analyzes a single audio file for specified metrics. Use this for quick, individual file processing.

```python
from analyzeAudio import analyzeAudioFile

values = analyzeAudioFile('audio.wav', ['LUFS integrated', 'RMS peak'])
# Returns: [-23.5, -15.2]
```

--------------------------------

### Control worker process count with CPUlimit

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/configuration.md

The analyzeAudioListPathFilenames function accepts a CPUlimit parameter to control the number of worker processes used, defaulting to the system's available CPUs if not specified.

```python
from analyzeAudio import analyzeAudioListPathFilenames

# Use system default
rows = analyzeAudioListPathFilenames(files, aspects, CPUlimit=None)

# Use exactly 4 workers
rows = analyzeAudioListPathFilenames(files, aspects, CPUlimit=4)

# Use all available CPUs
import os
rows = analyzeAudioListPathFilenames(files, aspects, CPUlimit=os.cpu_count())
```

--------------------------------

### Register New Audio Analyzer

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Use the `@registrationAudioAspect` decorator to register a new analyzer function with a specified aspect name.

```python
@registrationAudioAspect(aspectName: str)
def analyzer(pathFilename: str | PathLike[Any]) -> float | None:
    # Analyzer implementation
    pass
```

--------------------------------

### Compare Two Audio Files

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Analyzers for comparing two audio files using metrics like Peak SNR and SDR. These functions take paths to two audio files as input.

```python
def analyzer(pathFilenameAlfa: str | PathLike[Any], pathFilenameBeta: str | PathLike[Any]) -> float | None
```

--------------------------------

### analyzeSpectralFlatness

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the spectral flatness of an audio signal using Librosa, returning the full spectral flatness time-series.

```APIDOC
## analyzeSpectralFlatness analyzeSpectralFlatness()

### Description
Analyzes the spectral flatness of an audio signal using Librosa, returning the full spectral flatness time-series.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Spectrogram input types and full signatures are detailed in the source documentation.

### Request Example
```python
# Example usage (assuming analyzeSpectralFlatness is imported)
# spectral_flatness_series = analyzeSpectralFlatness()
```

### Response
#### Success Response
- Returns the spectral flatness time-series as a numerical array.

#### Response Example
```json
[0.1, 0.12, 0.09, 0.11]
```
```

--------------------------------

### analyzeTempogram

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/INDEX.md

Analyzes the tempo of an audio signal using Librosa, returning the full time-series tempogram.

```APIDOC
## analyzeTempogram analyzeTempogram()

### Description
Analyzes the tempo of an audio signal using Librosa, returning the full time-series tempogram.

### Method
Not applicable (SDK function)

### Endpoint
Not applicable (SDK function)

### Parameters
Parameters are detailed in the source documentation, including input types and mathematical descriptions.

### Request Example
```python
# Example usage (assuming analyzeTempogram is imported)
# tempogram_data = analyzeTempogram()
```

### Response
#### Success Response
- Returns the full tempogram as a numerical array.

#### Response Example
```json
[[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]]
```
```

--------------------------------

### Registering a Custom Audio Aspect Analyzer

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/audioAspectsRegistry.md

Uses the 'registrationAudioAspect' decorator factory to register a custom analyzer function with a specific aspect name. The decorator automatically stores the analyzer and its parameters in the 'audioAspects' registry.

```python
from analyzeAudio.audioAspectsRegistry import registrationAudioAspect

@registrationAudioAspect('My Custom Metric')
def myAnalyzer(pathFilename: str | PathLike[Any]) -> float:
    # Compute and return a metric value
    return 42.0

# After decoration, the analyzer is registered:
# audioAspects['My Custom Metric'] = {
#     'analyzer': myAnalyzer,
#     'analyzerParameters': ['pathFilename']
# }
```

--------------------------------

### Analyze Tempogram from Waveform

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseWaveform.md

Computes a local autocorrelation tempogram from an audio waveform. Requires waveform data and its sample rate. Additional parameters can be passed to librosa.feature.tempogram.

```python
import numpy
import soundfile
from analyzeAudio.analyzersUseWaveform import analyzeTempogram

with soundfile.SoundFile('audio.wav') as f:
    sampleRate = f.samplerate
    waveform = f.read(dtype='float32').astype(numpy.float32).T

tempogram = analyzeTempogram(waveform, sampleRate)
```

--------------------------------

### Inspect Analyzer Registry

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Inspect the analyzer registry to list all available metrics and check the parameter requirements for a specific analyzer. This is useful for understanding the API's capabilities and configuration needs.

```python
from analyzeAudio import getListAvailableAudioAspects, audioAspects

# List all available metrics
metrics = getListAvailableAudioAspects()
print(f"Available metrics: {len(metrics)}")

# Check requirements for a specific analyzer
analyzer_info = audioAspects['SI-SDR mean']
print(f"Analyzer requires: {analyzer_info['analyzerParameters']}")
# Output: ['pathFilenameAlfa', 'pathFilenameBeta']
```

--------------------------------

### Save Measurements to TSV

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/README.md

Analyze audio files and save the measurements to a Tab-Separated Values (TSV) file. Requires specifying the output filename, the measurement rows, and the columns to include.

```python
from analyzeAudio import (
    analyzeAudioListPathFilenames,
    dataTabularTOpathFilenameDelimited,
)

listAspectNames = ["LUFS integrated", "true_peak maximum"]
rows = analyzeAudioListPathFilenames(["one.wav", "two.wav"], listAspectNames)

dataTabularTOpathFilenameDelimited(
    "measurements.tsv",
    rows,
    ["pathFilename", *listAspectNames],
)
```

--------------------------------

### Analyze SRMR with PyTorch Tensor

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-reference/analyzersUseTensor.md

Computes Speech-to-Reverberation Modulation Energy Ratio (SRMR) values from an audio tensor. Ensure the audio tensor and sample rate are correctly provided. The `pytorchOnCPU` parameter controls CPU execution path.

```python
import torch
from analyzeAudio.analyzersUseTensor import analyzeSRMR

# Create dummy audio tensor
audio_tensor = torch.randn(1, 16000)  # 1 second at 16 kHz
srmr_values = analyzeSRMR(audio_tensor, 16000, pytorchOnCPU=True)
mean_srmr = srmr_values.mean()
```

--------------------------------

### Analyze Single Audio File

Source: https://github.com/hunterhogan/analyzeaudio/blob/main/_autodocs/api-overview.md

Use `analyzeAudioFile` to analyze a single audio file. It takes the file path and a list of aspect names to compute. Returns a list of computed values.

```python
def analyzeAudioFile(pathFilename: str | PathLike[Any], listAspectNames: list[str]) -> list[str | float]
```