Forests Have Voices, and Data Have Owners: How Soundscape Monitoring Can Avoid Turning Indigenous Communities into Free Training Sets

Soundscape monitoring has become increasingly popular in international research and conservation. The reason is straightforward: devices left in a forest can record birdsong, insect calls, streams, wind, and other signs of biological activity over long periods. They generate large volumes of data that are sensitive to ecological change, at a lower cost than keeping researchers continuously on site. To a researcher, this is like giving the forest ears. For many communities, however, the question is not only “What did you hear?” but “What gives you the right to take these sounds away?” For Indigenous communities, a forest has never been an ownerless natural backdrop, and sound is not a neutral repository of raw material. Sound carries seasons, labor, knowledge of hunting grounds, ritual boundaries, culturally restricted knowledge, and the ways people and their environment recognize one another.

A Soundscape Is Not Free Material; It Is Part of Local Relationships

In many technological settings, sound is divided into files, waveforms, annotation fields, and model-training sets. That division is convenient for engineering, but it can sever local relationships at the very first step. For Indigenous communities, some sounds may belong to contexts of knowledge that only particular groups may approach at particular times and in particular places. Some environmental sounds are not simply natural signals; they are intertwined with memories of the land, ancestral experience, rhythms of work, and rules for everyday life. If a research team merely installs recorders, removes the data, uploads it to the cloud, and then releases it publicly or uses it to train a model, the activity is no longer only monitoring. It is a classic form of data extraction.

Some people ask: bird calls are not human speech, so how can sovereignty be involved? The question itself exposes a common blind spot in mainstream research. Data sovereignty has never concerned only human voices or personal information. It also concerns the places in which data are generated, the process of collection, authority over interpretation, and mechanisms for returning results. When data come from Indigenous territories and environments sustained through local daily life and community rules, the community has a right to help decide what may be recorded, what may not, what may be made public, what must remain for internal use, whether the data may later be used in a commercial model, and how the research results will return to the community.

The Myth of Neutral Technology Often Conceals Unequal Terms of Access

A sensor may be quiet, but that does not make it neutral. Many monitoring projects appear well intentioned: they aim to study biodiversity, protect habitat, accumulate evidence, or build sound-recognition models. Good intentions do not remove the need for procedural justice. If outside institutions control the funding, equipment, and publication while the community is treated merely as a provider of a site, the resulting knowledge and benefits will probably continue to flow outward. Researchers receive data, papers, models, and credit toward project metrics. The community is left only with the trace of having been monitored.

This pattern is familiar in work involving Indigenous traditional knowledge. Knowledge has often been misread not because it does not exist, but because interpretive authority lies elsewhere. Mainstream institutions are adept at translating local experience into familiar formats, yet may be unwilling to recognize a community’s authority over its rules, taboos, and contextual boundaries. If soundscape data follow the same path, the technology may be new, but the structure remains old.

Two-Eyed Seeing Is More Than Sending Community Materials to a Laboratory

To approach soundscape monitoring through Two-Eyed Seeing, it is not enough to place Indigenous elements in the introduction to a research project. Different knowledge systems must have genuine roles in method, interpretation, and governance. Scientific monitoring can produce long-term, measurable, and comparable data. Local knowledge can explain which sounds are meaningful in a particular season, place, or set of relationships; which contexts must not be disclosed casually; and which changes matter most to the community. Two-Eyed Seeing is not decoration. It requires a redistribution of power over who defines the question, identifies risk, interprets the results, and shares the benefits.

In other words, a soundscape project that does not include the local community as a co-designer from the stage of defining the research question can hardly claim to practice Two-Eyed Seeing. Indigenous communities are not suppliers of research material; they are co-holders of knowledge and co-governors of the project. This is a methodological requirement as well as an ethical position. Without the participation of local knowledge holders, many sounds cannot be interpreted correctly. A model may identify a species of bird without understanding how its seasonal appearance relates to local work and ritual. A sensor may record a shift in frequency without knowing why a device should not remain permanently in a particular area. Technology that lacks humility can easily mistake what it does not understand for comprehensive knowledge.

Discuss Consent Before Openness, and Return Before Sharing

Amid the movement for open science and data sharing, it is easy to treat “the more open, the better” as the morally superior position. That principle does not necessarily hold for soundscape data from Indigenous communities. Openness is not the goal in itself; legitimacy, consent, and return are prerequisites. Some data may be made public for conservation advocacy, some may be available to researchers under defined restrictions, and some should not leave the community because they involve sensitive locations, taboo contexts, or internal rules. This is not resistance to openness. It is governance.

A responsible soundscape project should therefore meet several basic conditions: advance notice and free, prior, and informed consent; joint decisions about where and when data will be collected; clear rules for ownership, access, and licensing; results returned in forms the community can understand and use, rather than only an academic paper in English; benefit sharing and boundaries on use agreed in advance if research or products create commercial value; and local participation in annotation and interpretation, so outside researchers do not monopolize the authority to name. These are not optional additions. They are the minimum conditions for avoiding the conversion of a community into a free training set.

Forest Monitoring Can Serve Communities, Not Just Academic Papers

Soundscape technology can be valuable; the question is how it is used. Designed well, it can become a tool of local governance. It can help a community observe long-term changes in streams and bird activity, compare ecological recovery after typhoons, support environmental education, build a community-owned database, and contribute to cultural documentation and youth capacity-building. But the project must begin with local needs rather than outside research metrics. If data only flow outward, the community remains merely an object of study. If the data stay in the community together with the capacity to use, interpret, and manage them, the technology can genuinely return value to the community.

For research involving Indigenous peoples around the world, the most important change now is not the installation of more devices but a reconsideration of the research relationship. A sensor may be physically small, but unequal power can still make it intrusive. A recording may be technically excellent, but without community consent, quality cannot substitute for legitimacy. Precision monitoring should identify not only a species of bird, but also who has the authority to speak for and make decisions about the data.

Letting the Forest Be Heard Must Not Silence the Community Again

We can expect technology to help protect forests, and we can hope that soundscape monitoring will contribute more to climate, ecology, and conservation work. But if the technology enters an Indigenous community focused only on downloading sound, labeling species, and publishing results while forgetting consent, taboos, and local sovereignty, it may ultimately protect not the forest but the knowledge ambitions of outside institutions.

Yuan Media AI does not advocate replacing Indigenous traditional knowledge—including its rules and taboos—with artificial intelligence. The aim is to help knowledge contexts that have long been ignored, misclassified, or spoken for by others regain their own place, name, and authority of interpretation. If soundscape monitoring is genuinely to become part of Two-Eyed Seeing, it must first learn one principle: forests have voices, and data have owners. Without that premise, even the most advanced recording equipment can become another quiet form of extraction.

Sources retained from the Chinese original

• Original source: Global Indigenous Data Alliance | CARE Principles for Indigenous Data Governance
• Original source: Local Contexts | Traditional Knowledge and Biocultural Labels
• Original source: IPBES | Indigenous and Local Knowledge