Why Use SonoBat?
Fine scale analysis, meticulous and robust data extraction,
true full-spectrum analysis, for the most accurate classification available.
The only North American classifier to not just accept full-spectrum data, but to also perform data extraction and analysis using full-spectrum data processing. SonoBat renders high resolution sonograms of each call pulse to support the SonoBat intelligent call trending algorithm to extract full call trends through low amplitude or noisy conditions inaccessible to power trending methods such as zero-crossing. Extracting more information content from your recordings means better results.
Learn more about the SonoBat approach to bat call analysis and classification.
Extensive and broadly representative full-spectrum library.
SonoBat bases its classifiers on the most robust, species-confirmed full-spectrum reference library from continental-wide meticulous tracking, recording, and interpreting bats in the field. A systematic project begun in 1991:
Learn more about the history and development of SonoBat.
Learn more about the SonoBat approach to recording and how to improve your own recording.
All species-known sequence and call samples that contribute to the library from the field recordings also undergo rigorous manual vetting for inclusion; for example, to exclude unrepresentative signals such as distorted recordings or interloping species in the file.
Classification engine that integrates vetting experience.
SonoBat species classification integrates quantitative machine learning with algorithms that incorporate three decades of expert acoustic classification experience to produce results that surpass machine learning alone. The SonoBat classification system replicates the experience and decision steps an expert in bat species recognition would do to recognize call context, fragments, ambiguous call types, etc., i.e., the things to include and exclude to arrive at a species decision for a recorded sequence of calls, or to leave it appropriately unidentified if not reaching a sufficient threshold of confidence, and thus reduce false-positive results (but provide a suggested classification to guide manual vetting).
Extensive user tools and utilities built by bat biologists to meet the needs for bat call analysis.
We use these tools for bat monitoring and research. To streamline and support our own work we have implemented many tools and operations to streamline bat data processing and solve the problems that arise with bat data work.
Learn more about basic SonoBat operations and tools.
Learn more about SonoBat file processing, tools, and methods.
Learn more about SonoBat NABat-specific tools and methods.
The SonoBat Vetting Table works interactively with the SonoBat viewer to streamline vetting and reviewing results.
The SonoBat Data Wizard provides utilities to attribute files with metadata from most detector types, attribute files with NABat required metadata, scrub files to eliminate noise (non-bat) files, batch edit filenames, batch edit metadata notes, batch remove metadata, batch rename metadata fields, smart parse longe duration recordings into files of individual bat pass events, and batch convert .wav files to .zc files.
SonoBat LIVE performs uncompromised SonoBat analysis and classification in real-time with scrubbing and metadata attributing to eliminate any post-processing chores to present vetting-ready files.
Automated species classification with SonoBat
SonoBat brings automated processing of recorded files and automated species classification– indispensable tools for long-term and large-scale survey and monitoring projects. For data integrity and optimal classification performance, SonoBat performs a number of intelligent signal quality assessments and data extraction routines to provide the most faithful and confident call data.
True full-spectrum data extraction
Full-spectrum data includes signal information at multiple frequencies; high and low amplitude. SonoBat takes the extra care to access this content from each call pulse to track the full call trend available in your data. Other North American classifiers (even those that accept full-spectrum data) use zero-crossing analysis that can only access the strongest frequency content.
High performance classification
SonoBat evaluates not just call types, but calls within context of a sequence of calls, knowing when to accept or reject calls to contribute to a decision, just as an expert vetter does. SonoBat classification recognizes call fragments or distorted signals to reject them from contributing to decisions to uphold quality control and reduce false positive results.
Large project data management
Tools for project management, analysis, vetting, export, and reporting. Customizable Vetting Table that interacts directly with the Viewer, multi-level sorting, add any desired metadata column, export customized tables or select from outputs sorted by any metadata, output night by night and by project site night by night, output NABat required format, etc.