NatureLM-Audio

NatureLM-Audio

A demo where you upload an audio file of animal sounds and ask questions about it in a chat. It runs the NatureLM-audio model and opens in a browser with nothing to install.

Made by EarthSpeciesProject on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: NatureLM-Audio is a demo Space for the NatureLM-audio model. You upload an audio file that holds animal vocalizations and ask questions about it in a chat. It is built for bioacoustic data, such as field recordings of animal calls.

Who it suits: Researchers and students who work with recordings of animals will get the most from it. The README names no customers, so the rows below are judgment from what the demo does.

Industry | Who uses it | How they use it | Example
Research | Bioacoustics researchers | Ask the model questions about recordings from the field | Uploading a recording and asking what it hears
Conservation | Wildlife monitoring teams | Get a quick read on a recording before detailed analysis | Checking a clip before sending it to an expert
Education | Biology teachers and students | Explore animal sounds by asking questions in plain language | Asking what a short clip contains
Software | Audio-model developers | See how a chat interface is built around an audio model | Reading app.py in the repository

You can skip it if you want to analyze speech or music, since the README is about animal vocalizations.

Where to use it: Use it in a browser tab on Hugging Face, or build a copy in Docker on your own computer.

• A web browser: open the Space page on Hugging Face, where it runs on a100-large hardware
• Docker on Windows, macOS or Linux: build and run a copy

Use case | Who | What they get back
Ask about an animal recording | Researchers | An answer in the chat about the audio you uploaded
Try the model without installing | Anyone with a browser | A working demo, with the model loading on first use
Run a Docker copy | Developers | The app on port 7860 of their own machine

How to set it up: Using the Space online needs nothing. A local copy needs Docker and Python with pip, because the repository builds from a Dockerfile.

• Docker
• Python with pip, for the hub package
• Port 7860 free on your machine

1. Install the Hugging Face hub package.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Build the Docker image.
5. Run the container with port 7860 published.

pip install -U huggingface_hub
hf download EarthSpeciesProject/NatureLM-Audio --repo-type space --local-dir NatureLM-Audio
cd NatureLM-Audio
docker build -t naturelm-audio .

The model named in the record has private at the end of its name, so a local copy may not be able to load it. The README says the app then runs in demo mode.

How to use it: Open the Space page, upload an audio file with animal sounds and type your question in the chat. The model loads on first use, which can take a few minutes, and later questions are faster. On a local copy, start the container and open port 7860 in your browser.

docker run -it -p 7860:7860 naturelm-audio

Tool | What it does
Audio upload | Takes the file that holds the animal vocalizations
Chat | The place where you ask questions about the audio
Demo mode | What the app runs if the model fails to load

Example: Upload a short recording from a forest and ask: "What animal sounds are in this clip?" The chat answers about the audio. The README gives no sample answers, so judge the reply by listening to the clip yourself.

For newcomers and experts: Start with a short clip you know well, such as a single bird call, and ask a simple question. You can then judge whether the answer is right. Expect a wait on your first question while the model loads.

The app loads the model lazily, on the first interaction rather than at start-up, to avoid timeouts on Hugging Face Spaces, and it keeps the model loaded afterward. The Space builds from a Dockerfile, listens on port 7860 and runs on a100-large hardware.

You upload a clip of animal sounds
You ask a question in the chat
The model loads on first use
It answers about the audio

Good to know:

• License: Apache-2.0
• Cost to run: no API key or paid service named, and the Space runs on a100-large hardware at Hugging Face
• Data it reaches: the audio and questions you send. The record names a logs dataset and a hub_logger.py file, so usage may be logged, but the README does not say what is saved
• Not for: sound other than animal vocalizations, going by the README, and local runs that rely on the named model, which may be unavailable

On Hugging Face, 13 November 2026

• Likes: 14
• Built with: docker
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/EarthSpeciesProject/NatureLM-Audio
• Files: https://huggingface.co/spaces/EarthSpeciesProject/NatureLM-Audio/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.