A California-based non-profit group called Earth Species Project (ESP) has a bold ambition: to decode non-human communication using a form of artificial intelligence (AI) called machine learning, and make all the knowhow publicly available, thereby deepening our connection with other living species and helping to protect them. The organisation, founded in 2017 with the help of major donors such as LinkedIn co-founder Reid Hoffman, published its first scientific paper last December. The goal is to unlock communication within our lifetimes.
ESP is not the first to attempt to decipher animal vocalisations using AI. Various researchers have used machine learning algorithms to analyse the sounds of pigs, rodents, dolphins, whales, and birds, among others. But ESP claims to have a unique approach that combines cutting-edge technology with open-source data and collaboration. The organisation hopes to create a Google Translate for the animal kingdom, where anyone can access and contribute to the knowledge of animal communication.
How does ESP use AI to decode animal sounds?
ESP uses a type of machine learning called unsupervised learning, which does not require human labels or annotations for the data. Instead, the algorithm learns by finding patterns and structures in the data itself. This way, ESP hopes to avoid imposing human biases or assumptions on the animal sounds, and instead let the animals speak for themselves.
ESP has developed a software tool called ESPRESSO (Earth Species Project Representation Engine for Sound Signal Objects), which can process large amounts of audio data and extract features that represent different aspects of the sound, such as pitch, timbre, duration, and frequency. These features are then used to create a high-dimensional representation of the sound, which can be compared with other sounds using mathematical methods.
ESP has applied ESPRESSO to various datasets of animal sounds, such as elephant rumbles, dolphin whistles, and bird songs. The results show that ESPRESSO can cluster similar sounds together and distinguish different types of sounds from each other. For example, ESPRESSO can separate elephant rumbles based on their context, such as mating, greeting, or alarm calls. ESPRESSO can also identify individual dolphins by their signature whistles, and classify bird songs by their species and dialects.
What are the challenges and limitations of ESP’s approach?
ESP acknowledges that decoding animal communication is not an easy task, and that there are many challenges and limitations to their approach. One of the main challenges is the lack of data. Many animal sounds are rare or difficult to record in their natural habitats, and some animals may use other modes of communication besides sound, such as gestures, body language, or chemical signals. ESP aims to overcome this challenge by collaborating with other researchers and organisations that have access to animal sound data, and by making their data and tools open-source and accessible to anyone who wants to contribute.
Another challenge is the complexity and diversity of animal communication. Different animals may have different ways of encoding meaning in their sounds, and some animals may have more sophisticated or flexible communication systems than others. For example, some animals may use syntax or grammar to combine sounds into complex sentences, while others may use simple calls or signals to convey basic information. ESP does not assume that all animals have language in the same way that humans do, but rather tries to understand how each animal communicates in its own terms.
A third challenge is the ethical and social implications of decoding animal communication. Some people may question the validity or usefulness of ESP’s project, or raise concerns about the potential risks or harms of interfering with animal communication or behaviour. ESP believes that decoding animal communication can have positive impacts on conservation, education, and empathy for other living beings, but also recognises the need for ethical guidelines and public dialogue on how to use this knowledge responsibly and respectfully.
What are the future plans and visions of ESP?
ESP is currently working on expanding its data collection and analysis capabilities, as well as developing new tools and applications for decoding animal communication. Some of the future plans include:
- Creating a global network of automated listening posts that can record and analyse animal sounds in real time
- Developing a smartphone app that can recognise and translate animal sounds using ESPRESSO
- Building interactive devices that can communicate with animals using their own sounds
- Exploring the possibility of creating interspecies art or music using animal sounds
ESP’s ultimate vision is to create a new paradigm of understanding and relating to other living species on Earth. By decoding animal communication, ESP hopes to reveal the richness and diversity of life on our planet, and foster a deeper appreciation and respect for all forms of intelligence and expression.