`

Details

Company Name: Suez
Date: November 2021
Technologies: Automatic Speech Recognition (ASR)

Suez is a leading corporation in the field of water and waste management in France and worldwide. Suez France launched an innovation project to reinvent the tools provided to its employees in order to improve the collected data quality. In this context, Smile and Vivoka worked together to voice-enable the mobile application used to create work reports.

SUEZ, expert in water and waste management in France

SUEZ devises innovative solutions to support its customers in the transition from a linear model that overconsumes resources to a circular economy that recycles and recovers them.

In France, SUEZ contributes to the attractiveness of the regions and the sustainable development of its customers by offering solutions and services tailored to the environmental and energy transition. 

The company is the result of decades of development. Originally in 1880, Suez was the company “Lyonnaise des Eaux et de l’Eclairage (SLEE)”. The brand SUEZ was then created in 2015 with the fusion of multiple partners, brands and subsidiaries.

 

YEARS

EMPLOYEES

CLIENTS

ACTIVITIES

Project’s challenge

Improving the collected data quality through voice recognition on technician’s devices

With an approval and partnership from the Digital Acceleration Department of Suez, speech recognition was selected as the go-to solution on which a prototype could be created. Different problems were identified regarding reporting creation. Gaps in writing skills, environment conditions and existing tool ergonomics were impacting the data quality when technicians were doing their reports.

Voice Synthesis Embedded Logo VDK Vivoka

Letters and Digits Recognition

A large majority of the content technicians need to report are references that contains letters and digits. They are used to identify specific devices and give metrics about their status. In order to provide value to the user, digits and letters need to be perfectly recognized by the speech recognition engine.

Devices icon

Voice Application's Reliability

Technicians from Suez are often working on particularly hard conditions, in remote places. With such a vast area of action, network quality and stability cannot be taken as granted. In order to comply with this criteria, embedded technologies are more than recommended to grant reliability and robustness to the application.

Project’s solution

Embedded voice recognition (ASR-based) inside professional mobile application for report creation

Voice Technologies

Automatic Speech Recognition engine based on Grammar creation to create the voice commands and tailor the recognized words for Suez industry-specific vocabulary. Each voice command is designed with multiple alternatives that cover the different ways users can express.

Focus on End-Users

+1000 technicians from Suez are the expected end-users of this voice update. The solution is adapted to the current mobile application used for reporting, this time adding voice commands to ease the creation of reports by giving smart tools with increased ergonomics.

Hardware Involved

Speech recognition technology involved in this project is embedded inside mobile devices that are carried by technicians. The technology runs on Android 6.0 (API 23) and above. It doesn’t require specific modification to the device, the native microphone is totally adapted.

Some examples of use cases that were developed…

#1 - Meter Replacement

1) User – The index of the deposited meter: “The index number is “xxxx“.
2) System – “Indication by the machine of the current meter number”: “Please confirm the number xxxx“.
User – “Yes/No” confirmation by the employee
> If “Yes“, the procedure continues.
> If “No“, “System” asks to spell out – User: the collaborator spells out.
Resume step 3.

3) User – Announcement of the new index of the installed meter: “The index of the installed meter is 0″.
User – Announcement if a non-return valve is present: “I have installed a non-return valve“.

– Possible text said by the User :
I changed a water meter, the index of the installed meter is 0
I have changed a water meter, the index of the removed meter is 34458 M3
I have just renewed the meter, the removal index is 7899 M3, I have installed a check valve

#2 - Metering Station Leakage

1) User – Indication of the part concerned by the leak: “The part to be repaired is “xxxx”.
(warning, the parts database must be filled in beforehand, the listing must be provided).

2) User – Indication of the replacement part: “The part “xxxx” has replaced the part “xxxx“.
> The possibility to indicate if the employee to use a pump is possible.

3) User – The index of the deposited meter: “The number of the index is “xxxx”.

4) System – “Indication by the machine of the current meter number“: “Please confirm the number xxxx“.
User – “Yes/No” confirmation by the employee
> If “Yes“, the procedure continues.
> If “No“, “System” asks to spell out – User: the collaborator spells out.
Resume step 3.

– Possible text said by the User :
“I have changed a water meter, the removal index is 34458 M3.”
“I fixed the leak, it was just a problem of a seal to renew at the meter.”
“I have just renewed the meter, the removal index is 7899 M3, I have installed a non-return valve.”

#3 - Re-connection of a Branch

1) User – The meter index on day D: “The index number is “xxxx“.

2) System – “Indication by the machine of the number of the current personnel number”: “Please confirm the number xxxx“.
User – “Yes/No” confirmation by the employee
> If “Yes“, the procedure continues.
> If “No“, “System” asks to spell out – User: the collaborator spells out.
Resume step 3.

3) System – The machine asks if a counter was in place: “Was a counter in place?
> If “Yes“, the procedure continues.
> If “No“, “System“.

4) User – The re-connection to the BAC or to the meter is given by the employee: “I have re-connected the BAC connection.

5) User – Procedure of the new meter installed (seen in Use Case 1 and 2)
> Possibility to add if the employee has used a pump to evacuate the water or not.

– Possible text said by the User :
“I changed a water meter, the removal index is 34458 M3.”

“The customer has water, I put the connection back on the BAC.”
“I have put the service line back into service, the meter reading is 12998 M3.”

A project resulting from a recent technological partnership

Vivoka is providing Smile with the Voice Development Kit and training the partner’s collaborators to master the tool. The VDK contains all the required resources and technologies to build any voice-based interface.