CI Trainer app offers additional practice opportunities for people with a cochlear implant

Author: Rene van der Wilk Reading time: 3 minutes (Translated from Dutch. Original article can be found here)

For people with a cochlear implant (CI), it is important during rehabilitation to practise speech recognition regularly. In practice, however, this is not always straightforward. Not everyone has access to a practice partner every day or for sufficient amount of time. For those living alone, this option may sometimes be entirely unavailable. The CI Trainer app has been developed as an additional training tool. The exercises progress from individual words to full sentences, with background noise gradually introduced to simulate listening in more challenging environments. According to developer Wicher Donga of the company Enlightenment School, the use of AI and randomly generated word and sentence combinations makes the training less predictable and therefore more challenging.

CI Trainer app supports speech recognition training for cochlear implant users

Independent practice as a supplement to a training partner

CI Trainer was developed based on the creator’s own practical experience while undergoing a CI rehabilitation process. According to him, the idea arose due to limited practice opportunities within his immediate environment. A small social network or a partner who speaks a different language can also make regular practice more difficult.

Wicher Donga explains that he also evaluated existing training software from CI manufacturers. A key drawback, in his view, is that exercises can quickly become predictable, which may reduce their effectiveness. CI Trainer aims to address this by offering less repetitive training.

Progression from word pairs to sentences

The training consists of ten levels and begins with simple word pairs. The user selects the word that best matches what they hear. At later stages, the focus shifts to sentences, allowing the exercises to better reflect everyday communication.

The level of difficulty is automatically adjusted based on the user’s performance. A series of correct answers leads to more complex words, longer sentences, and more challenging listening conditions. These adjustments are made gradually, without requiring the user to manually select a new level.

Training with noise and unpredictable sentences

A key component of the training is listening in noise. The app uses broadband noise. According to the developer, the app itself determines whether and when noise is added, as well as in what form. The user has no direct control over this. This choice is deliberately linked to the scoring system, which automatically adjusts the level of difficulty.

Notably, the sentences are often constructed in a semantically illogical way. This reduces so-called semantic redundancy. As a result, users cannot rely as much on context or expectation and must depend more heavily on recognising actual speech sounds.

In addition, the CI Trainer app uses OpenAI to continuously generate new word and sentence combinations. According to the developer, this makes the exercises less predictable than fixed training sequences.

Voice selection enables targeted training

Users can choose whether speech is delivered by a male or female voice. In some languages, a child’s voice is also available. According to the developer, this allows for more targeted training, as different voices have distinct acoustic characteristics and therefore present different listening challenges.

Adaptive training without visible level jumps

The CI Trainer app continuously adapts to the user’s level. This is achieved through subtle changes, such as more complex vocabulary, longer sentences, softer speech, or increased background noise. Users can also manually select a level. If this level proves unsuitable, the app automatically corrects it.

According to the developer, this also makes the app suitable for tracking progress, for example within a speech therapy practice.

One-time purchase with separate OpenAI usage

CI Trainer costs €49.99 and is available in all supported languages in the App Store as a one-time purchase. In addition, the app uses OpenAI to generate training material. According to the developer, with daily use of around thirty minutes, these costs remain limited to a few euros, depending on the difficulty level. Words are quicker and cheaper to generate than sentences.

Wicher Donga states that he has deliberately chosen to let users manage this usage themselves, avoiding the need for a subscription and preventing unnecessary recurring costs.

Broader applicability beyond cochlear implants

Despite its name, the app is not limited to cochlear implant users. According to the developer, it can also be used by people with hearing aids or bone conduction systems. The focus remains on training speech recognition, regardless of the type of hearing device.

The developer of the CI Trainer app is also known for the HearingScan Pro app, which enables hearing tests via Mac, iPad, and iPhone, and for the Hearing Loss Estimator, a tool based on ISO models that calculates the risk of hearing loss due to sound exposure. A separate article on that application will follow. In addition, Wicher Donga develops other digital solutions. The CI Trainer app specifically focuses on speech recognition training during hearing rehabilitation.

A practical addition during rehabilitation

CI Trainer is intended as a supplement to existing rehabilitation, not as a replacement for practising with a partner or receiving professional guidance. For individuals who struggle to achieve sufficient practice hours, the app can provide an additional opportunity to work independently on speech recognition in various listening environments.

To download the CI Trainer app in the App Store, click here.

For additional information, click here.

Questions about the CI Trainer app? Click here to get in touch.