aleksipirttimaa/sm_exporter

Internet radio speech music discrimination using libopus and ffmpeg.

sm_exporter

Internet radio speech music discrimination using libopus and ffmpeg.

This script wraps ffmpeg and the analysis executable and exports the results as Prometheus metrics.

[icecast stream]  ->  [ffmpeg]  ->  [opus_sm]  ->  [prometheus]
                      ^^^^^^^^^^^^^^^^^^^^^^^  (sm_exporter scope)

You'll need a compiled opus_sm_demo executable and ffmpeg installed. This script forks them using submodule.

Metrics

Music probability

P("Content is music")

# HELP sm_music_probability Probability of music in stream
# TYPE sm_music_probability gauge
sm_music_probability{url="..."} 0.999795

Result instrumentation

This is a histogram measuring time between analysis results, you can use it to select an appropriate scrape_interval.

# HELP sm_analysis_result_s Instrumenting analysis executable output
# TYPE sm_analysis_result_s histogram
sm_analysis_s_bucket{le="0.005",url="..."} 26.0
  ...
sm_analysis_result_s_bucket{le="+Inf",url="..."} 791.0
sm_analysis_result_s_count{url="..."} 791.0
sm_analysis_result_s_sum{url="..."} 156.07464370007

Stream retries

This counter is incremented when ffmpeg or opus_sm_demo exit, resulting in them being reinitialized.

# HELP sm_stream_retries_total Incremented on retry
# TYPE sm_stream_retries_total counter
sm_stream_retries_total{url="..."} 0.0

opus_sm

You can find a fork of opus_sm that adds support for streaming here:

https://github.com/aleksipirttimaa/opus_sm

The build instructions in the aforementioned repository work just fine. On Debian you'll need to install some packages:

sudp apt install build-essential automake libtool

Just to recap building:

./autogen.sh
./configure
make

the resulting binary is ./opus_sm_demo

Running

1. Make sure you have opus_sm_demo and ffmpeg

which opus_sm_demo && which ffmpeg

2. Install pipenv, which is used for dependency management

python -m pip install --user pipenv

3. Create virtualenv and install dependencies

python -m pipenv install

4. run src/main.py

python -m pipenv run python src/main.py https://stream.example.com/example.m3u8

Add to prometheus

Add to your prometheus.yml:

scrape_configs:
- job_name: 'speechmusic'
    scrape_interval: 1s
    static_configs:
    - targets: ['localhost:9069']

A good scrape_interval depends on the incoming stream transport and encoding, as analysis can only be performed after an entire frame of the encoded audio has been received.