Menu

AI for complex signals, built on high-quality labeled data.

Ezako creates large datasets and specialized AI algorithms, turning complex signals into actionable intelligence.

Request a demoDiscover Upalgo
Request a demo
  • whistle00:01.0to00:03.0, 6.4 to 18.9 kHz
  • click train00:04.2to00:05.5, 1.8 to 23.7 kHz
  • sonar ping00:07.3to00:08.2, 7.2 to 7.8 kHz
A hydrophone signal computed for this page: 10 seconds, 0 to 24 kHz, three events labelled.

Offerings

Detection AI, datasets and expert annotation

  1. AI Algorithms

    Automatic detection and pre-labeling of vessel passages, acoustic signatures, biological sources and anomalies, validated by analysts.

  2. Datasets

    Large-scale labeled datasets, available off the shelf or built to specification, with rigorous versioning and traceability.

  3. Labeling Services

    Expert annotation of acoustic, DAS, radio-frequency and time-series data by Ezako specialists, on your premises or at Ezako, with strict confidentiality.

Software

The Upalgo suite

One labeling application in three editions, and the server that holds their recordings, their labels and the work of their teams.

  • Upalgo Labeling Sound: a spectrogram from 0 to 24 kHz with five boxed whistles, one of them selected, the list of events on the left and the selected event’s label, duration and bandwidth on the right.
    Upalgo Labeling Sound: a spectrogram from 0 to 24 kHz with five boxed whistles, one of them selected, the list of events on the left and the selected event’s label, duration and bandwidth on the right.

    Upalgo Labeling

    Sound

    Labeling of acoustic events on the spectrogram, for recordings of fifteen hours and more.

  • Upalgo Labeling DAS: the time–distance waterfall of a cable of six kilometres with a labeled vessel passage, the frequency–wavenumber spectrum of the selected window on the right, and the spectrogram of one channel underneath.
    Upalgo Labeling DAS: the time–distance waterfall of a cable of six kilometres with a labeled vessel passage, the frequency–wavenumber spectrum of the selected window on the right, and the spectrogram of one channel underneath.

    Upalgo Labeling

    DAS

    Labeling of distributed acoustic sensing data, in time and along the cable.

  • UpalgoDB: the list of labeling and validation tasks, each with its type, status, priority, assignee, and the files and labels it covers.
    UpalgoDB: the list of labeling and validation tasks, each with its type, status, priority, assignee, and the files and labels it covers.

    Server

    UpalgoDB

    The central server for recordings, labels and labeling teams, designed for petabyte-scale archives.

  • Upalgo Labeling Timeseries: four sensors of a turbofan test bench drawn one above the other across six engines run to failure, the nominal, onset and end-of-life ranges coloured on the chart, one range selected, the list of columns, labels and events on the left, and the values under the cursor.
    Upalgo Labeling Timeseries: four sensors of a turbofan test bench drawn one above the other across six engines run to failure, the nominal, onset and end-of-life ranges coloured on the chart, one range selected, the list of columns, labels and events on the left, and the values under the cursor.

    Upalgo Labeling

    Timeseries

    Labeling of sensor time series, with the search for similar patterns across series.

Sectors

Fields of application

  1. Naval and underwater

    Labeling of vessel passages, sonar emissions and biological sources on hydrophone and array recordings, with AIS tracks alongside the signal, and deep learning for underwater platforms.

  2. Distributed acoustic sensing

    Events located in time and along the cable on distributed acoustic sensing data, across thousands of channels.

  3. Signal intelligence

    Detection and labeling of radio-frequency emissions in time and frequency.

  4. Aerospace

    A published method to detect abnormal observations in engine test signals, and the detection of anomalies in satellite telemetry.

  5. Machines and installations

    The same methods apply to the sensors of machines, production lines and fleets of equipment.

Designed for sensitive environments

  • On-premises and offline

    Deployed on the customer’s infrastructure, including air-gapped networks. No data leaves the site.

  • Human validation

    Models pre-label, analysts decide. Every label records its author.

  • Full traceability

    Every change is auditable and reversible. Dataset versions are immutable.

How Upalgo is deployed and secured

They work with us

  • ThalesThales
  • Naval GroupNaval Group
  • MBDAMBDA
  • SafranSafran

See more references

See what Ezako can do with your data

Describe your signals and your requirements to arrange a demonstration.

Talk to the team