Talk2Nature FIELD COMPANIONYOUR SOUND NOTEBOOK
A moment worth noticing.
Open a sound. Notice what happened. Keep it together.
No account or upload. No species identification or animal translation.
Need to record first? Start listening →
Which files can I open?
Open a WAV sound (up to 2 minutes / 25 MB) or a Talk2Nature notebook (.t2n, under 28 MB). A notebook brings back its sound and observations together. MP3 and M4A are not supported yet. For a whole Station session, open the original ZIP in Compare—no unzipping needed. To annotate one clip, use “Annotate a clip from a saved session” below. Use recordings you have permission to inspect.
Annotate a clip from a saved session
Choose the original Station Save session ZIP (under 27 MB). We check the saved sound hashes and let you pick one clip without unzipping. Nothing is uploaded. The notebook will also carry a copy of the session journal, including its notes and declarations; it is not encrypted, so review it before sharing.
Save notebook downloads your sound and observations together. Until then, work stays only in this tab. The example uses invented tones; it does not access your microphone.
New here? Try the example. Your microphone stays off.
02 / INSPECT THE SOUND
Recording
Who was there? Not sure yet
Describe what you saw. Leave the animal unknown when unsure.
These details stay with this session. Save and discard the session before choosing another animal.
Your observation, not an automatic identification. Use no personal names or exact locations. Example sessions keep these as practice labels.
Sound details & level check
Drag across the waveform to mark an interval. On any device, you can also enter exact start and end times below.
Playback is only for your own review. Use headphones away from animals; do not broadcast recordings to them.
Advanced · identity & recording details
Use aliases, not personal names or sensitive locations. These fields are included in exported labels. The original filename is omitted.
04 / KEEP YOUR MOMENT
Your notebook.
No events yet.
One file with your sound and observations. Reopen it here next time. Stored on your device; no account needed.
Advanced · research formats & older labels
Open the same WAV first. Imports must match its checksum. JSON preserves the complete record; CSV is for inspection. These exports are preliminary annotations, not an approved training dataset.
Research exchange · experimental
Export separate AOEF 1.1.0 dataset and annotation-project files for a research annotation workflow. Add at least one event. The package keeps your unchanged WAV and a Talk2Nature provenance sidecar. The exact current export passed a synthetic Whombat 0.9.0 desktop-UI import with an invented WAV. This verifies the tested format path only; real-recording behavior and scientific suitability remain untested.
One-way export only. Human-entered labels stay separate from model predictions. The unencrypted sidecar preserves IDs, notes, optional animal context and Station source journal; these may identify people, animals or places. Inspect every file before sharing and keep the original notebook.
In Whombat, import dataset.aoef.json first with the extracted folder as audio directory, then import annotation-project.aoef.json. Official import guide ↗
Continue in Audacity
Take your sound and timed labels into Audacity for closer inspection. Add at least one event, then download the package and unzip it. Open recording.wav in Audacity and import labels.txt.
Includes the unchanged WAV, complete notes and a conversion report. Label times are rounded to six decimals; context and uncertainty remain display text. Edits in Audacity do not update this notebook. Keep your notebook to reopen here.
FROM LABELS TO EVIDENCE
Make it reproducible.
The companion Python tool checks exported labels and reports coverage, unknown context and overlapping events. Passing those checks does not verify permissions or establish meaning.
Read the workflow and run the quality report ↗
Help review this tool · Try the guided study demo · Privacy and project status