Home

Requirements

Python versions 3.8 and 3.9 are supported

Installation

Basic installation:

pip install hume

WebSocket and streaming features can be enabled with:

pip install "hume[stream]"

Basic Usage

Jupyter example notebooks can be found in the Python SDK GitHub repo.

Submit a new batch job

Note: Your personal API key can be found in the profile section of beta.hume.ai

from hume import HumeBatchClient
from hume.models.config import FaceConfig
from hume.models.config import ProsodyConfig

client = HumeBatchClient("<your-api-key>")
urls = ["https://storage.googleapis.com/hume-test-data/video/armisen-clip.mp4"]
configs = [FaceConfig(identify_faces=True), ProsodyConfig()]
job = client.submit_job(urls, configs)

print(job)
print("Running...")

job.await_complete()
job.download_predictions("predictions.json")
print("Predictions downloaded to predictions.json")

job.download_artifacts("artifacts.zip")
print("Artifacts downloaded to artifacts.zip")

Note: You can also supply a local filepath when submitting a batch job. Check it out in a Jupyter notebook here.

Rehydrate a batch job from a job ID

from hume import HumeBatchClient

client = HumeBatchClient("<your-api-key>")

job_id = "<your-job-id>"
job = client.get_job(job_id)

print(job)

Stream predictions over a WebSocket

Note: pip install "hume[stream]" is required to use WebSocket features

import asyncio

from hume import HumeStreamClient
from hume.models.config import BurstConfig
from hume.models.config import ProsodyConfig

async def main():
    client = HumeStreamClient("<your-api-key>")
    configs = [BurstConfig(), ProsodyConfig()]
    async with client.connect(configs) as socket:
        result = await socket.send_file("<your-audio-filepath>")
        print(result)

asyncio.run(main())

Other Resources

• Hume AI Homepage
• Platform Documentation
• API Reference

Support

The Python SDK is open source! More details can be found on GitHub.

If you've found a bug with this SDK please open an issue!