kaira.metrics.signal.SignalToNoiseRatio

Inheritance diagram of SignalToNoiseRatio — Inheritance diagram for SignalToNoiseRatio

class kaira.metrics.signal.SignalToNoiseRatio(name: str | None = None, mode: str = 'db', *args: Any, **kwargs: Any)[source]

Bases: BaseMetric

Signal-to-Noise Ratio (SNR) metric.

SNR measures the ratio of signal power to noise power, often expressed in decibels (dB). Higher values indicate better signal quality. It’s a fundamental metric in signal processing and communications [Goldsmith, 2005] [Sklar, 2001].

mode

Output mode - “db” for decibels or “linear” for linear ratio.

Type:: str

Methods

`__init__`	Initialize the SNR metric.
`compute_with_stats`	Compute SNR with mean and standard deviation across batches.
`forward`	Compute the Signal-to-Noise Ratio (SNR).
`reset`	Reset accumulated statistics.

Examples using `kaira.metrics.signal.SignalToNoiseRatio`

Simulating AWGN Channels with Kaira

Simulating AWGN Channels with Kaira

Composite Metrics

Composite Metrics

Metrics Registry

Metrics Registry

Signal and Error Rate Metrics

Signal and Error Rate Metrics

__init__(name: str | None = None, mode: str = 'db', *args: Any, **kwargs: Any)[source]

Initialize the SNR metric.

Parameters:

name (Optional[str]) – Optional name for the metric
mode (str) – Output mode - “db” for decibels or “linear” for linear ratio
*args – Variable length argument list passed to the base class.
**kwargs – Arbitrary keyword arguments passed to the base class.

forward(x: Tensor, y: Tensor, *args: Any, **kwargs: Any) → Tensor[source]

Compute the Signal-to-Noise Ratio (SNR).

Parameters:

x (Tensor) – The original (clean) signal tensor.
y (Tensor) – The noisy signal tensor.
*args – Variable length argument list (unused).
**kwargs – Arbitrary keyword arguments (unused).

Returns:

The computed SNR value(s). If input is batched, returns SNR per batch element.

Return type:

Tensor

compute_with_stats(x: Tensor, y: Tensor, *args: Any, **kwargs: Any) → Tuple[Tensor, Tensor][source]

Compute SNR with mean and standard deviation across batches.

Parameters:

x (Tensor) – The original (clean) signal tensor (batched).
y (Tensor) – The noisy signal tensor (batched).
*args – Variable length argument list (unused).
**kwargs – Arbitrary keyword arguments (unused).

Returns:

Mean and standard deviation of the SNR values across the batch.

Return type:

Tuple[Tensor, Tensor]

reset() → None[source]

Reset accumulated statistics.

For SNR, there are no accumulated statistics to reset as it’s a direct computation.

get_expected_args(): Return expected argument names for the metric.