We have developed our AI speech recognition solution optimized for the Hungarian language

2019/05/02

Speech recognition solutions based on artificial intelligence open up many untapped potential in all business sectors. Programs recognizing the English language are spreading explosively, but the Hungarian language has been pushed to the background so far. Until now! Get familiar with our own speech recognition solution based on deep-learning, optimized to Hungarian!

AI speech recognition

Various speech and voice recognition systems are of the utmost importance of the possibilities offered by artificial intelligence as they have many practical applications. With the quick spread of deep neural networks, there has been significant development in speech recognition, therefor well-known voice-controlled assistants such as Google Home, Amazon Alexa and other tools working with voice recognition are becoming more and more widespread.

Limitations and difficulties

Although the first speech recognition solutions appeared several decades ago, voice recognition has proved to be quite a complicated task and the algorithms used in the past have worked with rather poor accuracy. The spread of deep learning has been a major innovation in speech recognition, but it still has to overcome many challenges: the program must filter out background noises and it has to cope with the complexity of the human language. This includes different speech styles, pitch and the speed of the speaker. The problem is further complicated by the complexity of the Hungarian language. The diverse use of suffixes, the word-formation methods dissimilar to other languages, and the unstructured word order are all factors that cause major obstacles when teaching an artificial intelligence to Hungarian. In addition to these, the amount of data that can be used for training is significantly smaller than, for example, in the case of English-speaking AIs.

Our own speech recognition AI solution

By overcoming the above-described difficulties, we managed to create a high-precision speech recognition system with the help of deep neural networks, which is available on Hungarian and English language as well.

The great advantage of our solution is that the system can be implemented immediately with the use of general teaching material and can be further optimized for different professional languages, thus recognizing and describing the specific terms of the given field with a higher accuracy. Thanks to this, our artificial intelligence can work even more accurately adjusted to individual needs. For example, if we supplement the data used for training with medical vocabulary, it will be able to complete clinical documentation. But we can also think about recording IT, legal or technical texts - the possibilities are endless.

In contrast to dictation software solutions that use older technologies, it is not necessary to talk slowly to get precise speech recognition results, our program follows the standard speech rate. Hence, this AI solution enables real-time dictation, and the created text can be searched and analysed.

Try our AI solution in action!

 

 

 

 

 

 
 
Could you make a great use of artificial intelligence in your own company? Learn more about our AI solutions!
 
 

Possibilities of using a speech recognition artificial intelligence

 

Searching in audio and video files

Our AI solution can be used effectively to create subtitles for audio and video files. It also makes it easy to search and categorize the sounds of a huge database, so you don't have to seek for a single term for hours in different audio and video materials. Get rid of monotonous tasks!

 

Voice control

The ability to control smart devices with our voice is becoming increasingly important in our lives. Deep learning-based speech recognition can help with the management of voice-controlled assistants and other robots in English or Hungarian, or a customized voice management system can be developed for devices that do not have this function yet.

 

Transcription

Overcoming the slowness of typing can greatly increase productivity and creativity. The use of a well-prepared artificial intelligence for dictation can bring enormous benefits to those who spend several hours with typing on a daily basis. The great advantage of our solution is that it is capable of continuous text creation by following standard speech rate, so the desired text is created simultaneously with the speech.

Speech analysis (for customer services)

Combining speech recognition and text analysis features of artificial intelligence can be used to analyse human speech. This provides a great opportunity to measure customer satisfaction or employee efficiency. For customer services, you can define the terms and phrases that trigger the highest or lowest level of satisfaction, so you can maximize the performance of your service.

 

Our object recognition (image recognition) AI solution – Demo video