hyScore | media

Speech-to-Text

Transform Audio or Audio/Video in structured data.  

Solution

hyScore|media

Speech-to-Text (Audio/Video): The ability of a computer to recognize and understand human language. hyScore|media or “Speech-to-Text” is a process for converting speech/audio into digital text based on speech recognition.

The hyScore|media product helps us to categorize thousands of audio podcasts daily. If we have to listen to every single podcast by yourself, it would just be impossible to get direct insights and be as actual as can. With hyScore, we can automate the whole categorization process of the audio content and focus on other tasks. In direct comparison, it’s outstanding efficient and cost-effective instead of doing it manually!

Speech-to-Text

The sent audio file, for example a podcast, a video/audio track, an interview recording or any other audio media file, is transformed into machine readable text for further analysis. We perform the transcription of the audio/media file by our own technology or can use your own transcription data. The text can be provided in the analysis result as well.

Data Normalization

The keywords are provided in two forms: in the ‘surface form’ (how a user would hear the keyword) and in a ‘normalized’ version (data normalization), e.g. “Die Union, CDU, Christlich Demokratische Partei” (surface forms) will be in the normalized version always “Christlich Demokratische Partei” for all three spellings. 

Value Determination

Our sophisticated deep learning algorithms are extracting the most valuable keywords, entities, and categories scored and weighted from a transcribed audio/media file.

Sentiment Analysis

The default sentiment analyses and converts the expressed attitude or mood of the transcribed audio file into a general numeric value that indicates whether the text has a “positive” (max +1), “neutral” (0) or “negative” (max -1) tonality.

Languages

We’re multilingual. Our solution supports more than 20 languages by default. Such as Englisch, German, French, and other European languages or Hindi, Chinese, etc. – find a full list here! 

Categorization (default)

The solution returns weighted categories of the analyzed audio content in the result. Categories such as: Education, Politics, Sports, Entertainment and more. Over 400 default categories are available. A full list you’ll find in our documentation.

  IAB Taxonomy

The system returns by default the official IAB content taxonomy and IAB code in the latest version (v2) for the analyzed audio/media file in the hyScore|media response. The old IAB taxonomy (v1) is also supported. Be ready for online e.g. video or podcast marketing!

Custom Taxonomies

If you need a very individual taxonomy/categorization for your use case, we are able to train and customize the solution for you. If you need an individual categorization which is more detailed or focussed on a specific business – talk to us!

Platform independant

It’s simple. It is an Interface (API) working in the background. You can use hyScore|media for any kind of web- and mobile application or any other platform. As long you can send a request to our API handle the analysis response.

  Data Pass-back

Pass any kind of custom identifier (Podcast- or Video ID, cookie ID, Login or Product ID, Advertising or Device ID, any identifier you want) or custom data as a string through our system to enrich your data sets with our analysis results. For example: to enrich user profiles for your own or 3rd party DMP to create target group segments. 

Scalable & Fast

The analysis of an unknown audio/media file usually takes as long as the media file is. A minute is a minute. Our technology can speed up the analysis by factor 0.5, means instead of a (1) minute we can perform the analysis in 30 seconds. We’re talking in Real-time (direct). If you need End-of-Day or hourly processing, etc. it is possible.  

SSL secured

It’s simple. It is an Interface (API) working in the background. You can use hyScore|media for any kind of web- and mobile application or any other platform. As long you can send a request to our API handle the analysis response.

If you’ve any additional questions please get in touch with us!