BahasaKita Wants to Build the World's Best Voice Processing System for Indonesian

The information contained in a voice is actually very rich, but until now no one has realized its potential. This opportunity is being explored by BahasaKita, a company that focuses on developing information and technology-based products, especially speech and natural language processing. His vision is to be able to build the best sound processing system in the world, especially for Indonesian first.

BahasaKita operates under the umbrella of PT Bahasa Performance Utama, which was officially established in mid-2015. Oskar Riandi, who currently holds the position of Director, is the person who has played the most role in the birth of this BahasaKita service. Apart from Oskar, there are two other colleagues who also help from the financial side as angel investors.

Although it was only officially established as a company in the middle of 2015, but the process of developing a speech recognition system from BahasaKita actually has history which is much longer. Starting from when Oskar studied in Japan, until he decided to quit his career at BPPT in 2013 to fully enter the industry. The initial system of BahasaKita itself has been used for help people with disabilities to operate the computer.

Our language and developed services

"OurLanguage it is a company focused on developing natural language processing and speech processing technology products. For example, converting voice to text or vice versa, machine translation such as Google Translate, and we do have the ultimate goal of being able to translate speech to speech, so that it is multi-lingual. If we speak Indonesian, what comes out is English, [and vice versa] so as to reduce communication barriers between nations,” explained Oskar when met on the sidelines of the Big Data Conference in early December 2016.

The development of the Kita Kita service itself is also based on idealism and nationalism in mastering technology for the Indonesian language as national identity and asset. Mastery of this technology can give birth to independence in meeting other technological needs while increasing the nation's competitiveness against foreign products.

Speaking of products, Oskar also explained that currently there are several products developed and made by BahasaKita. Among them are Meeting Minutes, Keyboard Notes, KuTulis, EduGame, MoNik (Advertising Monitor), PoLIS (Proof of Life Information System), and Electronic Medical Records. However, the only thing that has really hit the market and is used by various government agencies or companies is the Minutes of Meeting.

Simply put, this Meeting Minutes has the ability to change the voice of the speaker into writing. The level of accuracy and speed of the translation is also very high because According to Oskar, there are more than 300 thousand words stored in the KitaKita system today.

In addition to the voice sourced from the direct conversation, the Meeting Minutes can also change the recorded voice of the conversation in an electronic file (WAV file) 16 kHz & 16 bit. The translation of the voice can later be directly corrected real-time and saved in paragraph form or just text for further modification.

Interestingly, Oskar also claims that his voice recognition system can distinguish the source of the voice from the speaker, be it person to person or gender.

The business model, the constraints faced, and the vision to be achieved

Oskar explained that currently his party is marketing the products it makes with a license sales model,on-premisesystem. After purchasing the product, the client basically owns the product for life, but there is annual technical support which can be extended later. Another model is through open API that can later be used by developers others to develop other applications.

Oskar said, “For this Meeting Minutes, [what is being sold] is in the form of a system, there is hardware and software, and there are also maintenance for adaptation. So, [can] adjust to the user's condition because we don't know what kind of room this user will have. […] Noisy, echoing, and so on.”

“We have the ability to adapt to the user's room. So that our system is not just a one-time sale and done. […] This perpetual license then there is one year technical support. […] The license is valid for life, technical supportIt can be extended,” continued Oskar.

In its development, Oskar also revealed that in fact there are obstacles that he is currently facing that come from the infrastructure side. This is what makes products like KuTulis and also the developed Notula Keyboard unable to take off smoothly.

Oskar explained, “There are infrastructure constraints that prevent us from being able to put [services] in cloud because this processing requires a very high computational infrastructure and powerful. How come I just found out, you can see who [currently] plays in speech recognition it's all giant companies like Google, IBM, Microsoft, Nuance, and others because cost to make this system possible on-cloud it was huge.”

However, Oskar also continues to conduct research on service development so that the system can continue to run and not crash cloud. It also attempts to store data offline on smartphones. In addition, Oskar also still dreams of being able to develop the Kita Kita system to become one of the best in the world, at least for Indonesian first.

“To be able to improve our performance, we need bigger data going forward. […] Our vision in the future is to continuously improve the performance of this system so that it can be the best in the world, especially for Indonesian first. Of course we are also open to collaboration with various parties. From our side we can prepare data [words from voice translation] for analytics[big data],” said Oskar.