Raven Pro

View any combination of waveform, spectrogram, spectrogram slice (single spectral frame), or selection (average) spectrum views.

Independent control of spectral window size and DFT size enables improved frequency measurement precision. Specify spectral window size by sample points, time units, or corresponding bandwidth. Choose DFT size directly or by corresponding frequency measurement precision. Variable overlap between spectral frames. Choice of six spectral window functions.

Raven’s powerful view linkage controls enable you to simultaneously view a signal at multiple time or frequency scales or at different spectral analysis resolutions. Link time and frequency scales so signals in different windows zoom and rescale together, facilitating visual comparisons between recordings.

Multi-line wraparound spectrogram views let you look at longer time periods while preserving a fine temporal viewing scale.

A workspace saves all information about Raven's state, including what signals are open, the size and placement of their windows, what views exist for each signal, selections in each signal, and linkages among views within and between signals. You can later retrieve a workspace to continue your work exactly where you left off.

Presets let you easily switch among frequently used configurations without having to specify numerous parameters for window layouts, color schemes, measurement sets, spectrograms, recorders, band filters, selection labels, and detectors.

A color scheme consists of a spectrogram colormap relating power values to display colors, and color specifications for elements within each view (such as selections, axes, backgrounds, title text, etc.). Choose from eight predefined colormaps, and edit the color for each element in a view.

Raven checks for software updates each time it is opened.

Paged sound windows enable fast and easy navigation and viewing of recordings of any length, consisting of any number of sound files (for example, a full year of continuous recording in 1-hour sound files).

Visually review long recordings (for example to search for specific sounds of interest) many times faster than real-time speed.

Open sound files saved in WAV, AIF, MP3, and FLAC formats.

Open sound files with any number of audio channels. Open all channels or a specified subset of channels.

Export images of signal views or sound windows via a variety of graphics file formats, or via the system clipboard for easy import into word-processing, presentation, and graphics programs.

More than 70 measurements are available for rectangular time-frequency selections around signals of interest. Selections can be created manually by drawing boxes on spectrograms or other views, or by Raven’s band-limited energy detector (see below). Selections measurements are shown in rows in a spreadsheet-like selection table. In addition to measurements computed from the audio signal, selection tables can include multiple user-defined annotations.

Selection tables can be saved as text files for easy import into statistics, spreadsheet, or other programs

Raven’s spectrogram correlation tool provides a quantitative measure of similarity between two sounds. Batch correlation generates a square correlation matrix containing similarity measures for all possible pairs of input sounds. Correlation matrices can be exported for use in in other programs, for example in cluster analysis or comparisons to other types of data.

Band-Limited Energy Detector (BLED)

Rapidly scan recordings to automatically create selections around events of specified durations in a user-defined frequency band that exceed the background noise level by an adjustable threshold.

Machine Learning Detector

We are thrilled to announce the latest release of Raven Pro which includes a major leap forward: Machine Learning is now in Raven!

In the new release, Raven introduces "Learning Detector" as a platform to allow user to integrate machine learning models as automatic detection and classification tools. Four models are included in this release as examples:

• BirdNET (3,000+ avian species)
• Katydid (30+ Panama katydid classes)
• Blue whale
• North Atlantic right whale

Raven users can train and add their own models. Documentation for the Learning Detector, including how to prepare your models to be used in Raven, can be found in
https://ravensoundsoftware.com/article-categories/learning-detector/

Raven’s Selection Review interface accelerates review of large numbers of selection generated by an automated detector by displaying a matrix of thumbnail images of selections. Multiple selections can then be annotated with a single keystroke, for example to label true versus false detections or to assign call type or species labels.

Compare the results of a detector run against a “truth” table created by a human. Raven automatically labels true positive events (TP) that the detector found, false positives (FP), and the false negatives (FN) that the detector missed. This feature facilitates quantitative assessment of detector performance.

Raven Pro supports audio data acquisition using the computer’s built-in sound card, NI-DAQ (National Instruments Data Acquisition hardware), or ASIO (Audio Stream In/Out) audio input devices. NI-DAQ devices provide input capabilities of up to 32 channels with an adjustable voltage range, and ASIO devices provide 24-bit audio sample size capabilities.

Raven displays waveform, spectrogram, and spectrogram slice views in real time during data acquisition.