AI Radio Streaming App Success
Matellio helped a startup (now public) company to achieve a critical product quality threshold by development of novel AI / ML techniques, thereby accelerating product launch and revenue generation through Apple and Google app stores.
01
The Challenge
A VC-backed startup developing an AI-enabled radio streaming app with uninterrupted, personalized and local AM/FM content had reached a plateau in improving the accuracy of its AI to detect and remove advertising or offer alternative advertising for local markets when the audio is sourced from a different region.
The client sought external expertise to increase its advertisement detection success rate in order to improve customer user experience, accelerate product launch, and enable revenue generation.
02
The Solution
Matellio ran two parallel research streams during development, one focused on tuning of the client’s existing model, the other focused on alternative approaches such as spectrographic analysis.
Analysis and modeling was based on a large dataset of existing audio samples and required developing multiple data analysis approaches and views.
Our AI experts analyzed pitch and cadence variations, key word patterns, and dominant channels (speaking versus background music) at granular audio sample sizes to identify novel audio classifiers that improved and increased detection of advertisements.
03
The Outcome
Prior to Matellio’s engagement, the client’s advertisement detection success rate was 90%. Through design and application of our novel modeling and analysis efforts, we increased the success rate to 98%, a critical threshold for customer adoption.
Through Matellio’s project services, the client was able to better meet customer expectations, instill confidence in VCs & private investors, and advance product development and release efforts.
Shortly after project completion, the client launched its freemium product in app stores (over 100K downloads) and is publicly traded on Nasdaq.
