1. Default Hop Size (API, SDK)

Cochl.Sense analyzes audio data in “window” units, where each window is a 1-second block of audio data. The window hop represents the time gap between consecutive windows, indicating the frequency of inference in seconds.

You can now control the hop size.

• When disabled, the Cochl.Sense predicts without overlapping intervals, using a 1-second interval, enabling real-time operation in environments with limited computing power.
• When enabled, it predicts with a 0.5-second overlapping interval, resulting in more accurate predictions.

The window hop is adjusted with WindowHop Enum.

• HOP_500ms (default)
• HOP_1s

import cochl.sense as sense

api_config = sense.APIConfig(
    window_hop=sense.WindowHop.HOP_1s,  # or sense.WindowHop.HOP_500ms
)
client = sense.FileClient(
    "YOUR_API_PROJECT_KEY",
    api_config=api_config,
)

2. Sensitivity Control (API, SDK)

Previously, the sensitivity of Cochl.Sense was a fixed value. The new version allows users to adjust the sensitivity so that it can be customized depending on target sounds and user scenarios. If you find that tags are not being detected accurately, try increasing the sensitivity. Conversely, if there are too many false detections, consider lowering the sensitivity.

• When disabled, it follows the default sensitivity of all tags.
• When enabled, it follows adjusted sensitivity levels from very high to very low.

import cochl.sense as sense

api_config = sense.APIConfig(
    sensitivity=sense.SensitivityConfig(
        # default sensitivity applied to all tags not specified in `by_tags`
        default=sense.SensitivityScale.LOW,
        by_tags={
            "Baby_cry": sense.SensitivityScale.VERY_LOW,
            "Gunshot":  sense.SensitivityScale.HIGH,
        },
    ),
)
client = sense.FileClient(
    "YOUR_API_PROJECT_KEY",
    api_config=api_config,
)

3. Result Summary (API, SDK)

• When disabled, it provides detailed results for each prediction.
• When enabled, it provides more concise results for each prediction.

NOTE: The result of stream feature does not support summarized format because it outputs its result in real-time.

4. Loudness Filtering (SDK-Stream Only)

Very small sounds are often not so meaningful compared to the sounds above certain loudness. The ‘loudness filtering’ function removes all the small sounds from the results so users can focus on meaningful sounds.

• When disabled, it provides all results without filtering.
• When enabled, it provides the results with unreliable results filtered out.