Audio Quality
Evaluates the perceptual quality of an audio file, assessing clarity, noise levels, and overall listenability.
Audio Quality checks how clear and listenable an audio file is, independent of what’s being said. Run it to catch noise, distortion, and other recording issues before they affect downstream processing.
What it does
Audio Quality is an LLM-as-Judge eval. It listens to the audio and scores its perceptual quality, covering clarity, background noise, and distortion.
Input
| Required Input | Type | Description |
|---|---|---|
input_audio | string | The file path or URL to the audio file to be evaluated |
Output
| Field | Type | Description |
|---|---|---|
| Result | score | Higher scores indicate better audio quality |
| Reason | string | A plain-language explanation of the audio quality assessment |
Run it from code
Call evaluate() with the template name and the eval’s required inputs. It returns the score and the reason.
Note
Before running: install the SDK and set FI_API_KEY / FI_SECRET_KEY. The model argument in the snippets is the evaluator model Future AGI uses to run the eval; turing_flash is a fast default.
from fi.evals import evaluate
result = evaluate(
"audio_quality",
input_audio="https://datasets-server.huggingface.co/assets/EarthSpeciesProject/NatureLM-audio-training/--/e98500754629b63dd8d2400c1a20798337da92f5/--/NatureLM-audio-training/train/0/audio/audio.wav",
model="turing_flash",
)
print(result.score)
print(result.reason)import { evaluate } from "@future-agi/ai-evaluation";
const result = await evaluate(
"audio_quality",
{
input_audio: "https://datasets-server.huggingface.co/assets/EarthSpeciesProject/NatureLM-audio-training/--/e98500754629b63dd8d2400c1a20798337da92f5/--/NatureLM-audio-training/train/0/audio/audio.wav"
},
{ modelName: "turing_flash" }
);
console.log(result);When to use
Run Audio Quality wherever an audio file’s recording condition matters on its own, separate from its content.
- Audio inputs and outputs, to catch clarity or noise issues before further processing
- Voice pipelines, to filter out recordings that are too degraded for reliable downstream use
- Quality gating for uploaded or generated audio before it reaches a listener
What to do when Audio Quality fails
If the audio quality score is lower than expected:
- Check for background noise or interference in the recording
- Verify the recording environment is appropriate (e.g., proper acoustic treatment)
- Ensure the microphone or recording device is of sufficient quality
- Consider using noise reduction techniques in post-processing
- Check for issues like clipping, distortion, or compression artifacts
- Verify the audio file format and bitrate are appropriate for the intended use
- Re-record in a more controlled environment if possible
Questions & Discussion