Service: Audio QC - EyevinnOSC/community GitHub Wiki
Audio QC (serviceId eyevinn-audio-qc) analyzes audio files for EBU R128 loudness compliance. In Eyevinn Open Source Cloud (OSC) it is a job: it runs the audio-qc command once and then finishes, rather than staying up as a service. License: MIT. Source: Eyevinn/audio-qc.
- Check audio loudness and quality against EBU R128 and produce a JSON compliance report.
- Read input from S3-compatible storage (AWS S3, MinIO) or from an HTTP/HLS URL, including audio inside video containers (MP4, MOV, MXF) and WAV.
- Upload the report to an S3 bucket, or print it to stdout (visible in the job logs) when no bucket is given.
- Select a specific audio stream and apply broadcast or music analysis modes through command-line flags.
Audio QC has no OSC service dependencies. If you read from or write to S3 storage, create a bucket first (for example with MinIO, minio-minio) and store the credentials as service secrets with create-service-secret (serviceId: "eyevinn-audio-qc"), for example accesskeyid and secretaccesskey.
Create the job with create-service-instance and serviceId: "eyevinn-audio-qc":
{
"name": "audiocheck1",
"s3AccessKeyId": "{{secrets.accesskeyid}}",
"s3SecretAccessKey": "{{secrets.secretaccesskey}}",
"s3EndpointUrl": "https://<tenant>-<minio-name>.minio-minio.auto.prod-se.osaas.io",
"cmdLineArgs": "analyze s3://input/video.mp4 --json --s3-bucket output --s3-key reports/audiocheck1/compliance.json"
}Follow the run with get-instance-status and get-logs-for-instance. The same job can be created with the OSC CLI:
% export OSC_ACCESS_TOKEN=<your-personal-access-token>
% npx -y @osaas/cli create eyevinn-audio-qc audiocheck1 \
-o s3AccessKeyId="{{secrets.accesskeyid}}" \
-o s3SecretAccessKey="{{secrets.secretaccesskey}}" \
-o s3EndpointUrl="https://<tenant>-<minio-name>.minio-minio.auto.prod-se.osaas.io" \
-o cmdLineArgs="analyze s3://input/video.mp4 --json --s3-bucket output --s3-key reports/audiocheck1/compliance.json"| Option | Required | Type | Meaning |
|---|---|---|---|
| name | Yes | string | Name of the job. Alphanumeric/word characters only (^\w+$), at most 20 characters. |
| cmdLineArgs | Yes | string | The full command line passed to the audio-qc CLI: the analyze command, the input file or URL, and flags such as --verbose, --json, --s3-bucket, --s3-key, --audio-stream, --broadcast, --music. Example: analyze https://example.com/video.mxf --verbose --json --s3-bucket reports. |
| s3AccessKeyId | No | string, sensitive | Access key ID for the S3-compatible storage used to download input or upload reports. Corresponds to S3_ACCESS_KEY_ID; AWS_ACCESS_KEY_ID is the fallback. |
| s3SecretAccessKey | No | string, sensitive | Secret access key for the same storage. Corresponds to S3_SECRET_ACCESS_KEY, which takes priority over AWS_SECRET_ACCESS_KEY. |
| awsRegion | No | string | Region of the S3 bucket, or the region setting for an S3-compatible service. Corresponds to S3_REGION, falling back to AWS_REGION. |
| s3EndpointUrl | No | string | Custom S3-compatible endpoint URL for services other than AWS S3, such as MinIO (for example https://<minio-host>). Corresponds to S3_ENDPOINT. |
| awsSessionToken | No | string, sensitive | Session token for temporary AWS credentials (for example assumed IAM roles). Corresponds to S3_SESSION_TOKEN. |
Sensitive options should be passed as {{secrets.NAME}}; a reference must be the whole field value, and a missing secret gives no immediate error, so confirm the secret exists. The S3 options can be omitted when the input is an HTTP URL and no report is uploaded.
Example cmdLineArgs values:
- Specific audio stream with verbose output:
analyze s3://input/multichannel.mov --audio-stream 1 --verbose --json --s3-bucket output --s3-key reports/audiocheck2/detailed.json - Broadcast standards:
analyze s3://input/broadcast.mxf --broadcast --json --s3-bucket output --s3-key reports/audiocheck3/broadcast-compliance.json - HLS stream over HTTP, report to stdout (no S3 options needed):
analyze https://vod.demo.osaas.io/osc-reel-9dd16/index.m3u8 --json - Music content:
analyze s3://input/music.wav --music --json --s3-bucket output --s3-key reports/audiocheck5/music-analysis.json
The JSON report contains integrated loudness (LUFS), loudness range (LU), true peak (dBTP), a compliance status, and details of any violations.
- Check that the input format is supported (MP4, MOV, MXF, WAV and similar).
- The S3 credentials need write access to the report bucket; HTTP inputs must be publicly accessible.