Voice Search: How to Find Spoken Keywords and Phrases in Audio Files

“I have hundreds, thousands of audio files from meetings (lectures, news, call centre recordings, podcasts). Is there software to search for particular words in these audio files?” The common answer to this question is transcribing audio files via automatic speech recognition and then searching words or phrases within the text output. However, readers who tried this approach might have experienced some drawbacks. Automatic speech recognition (ASR) struggle to with proper nouns such as brand, product or individual names. For example, an ASR engine would transcribe Toronto’s famous Yonge Street as Young Street.

If you’re looking for technology similar to the Google search engine that enables keyword search by crawling audio files instead of websites, you need a Speech-to-Index engine.

Why acoustic-only speech indexing?

Structuring unstructured audio and video data for monitoring, compliance and analysis will help enterprises minimize their risks and monetize this large data. Picovoice team has gathered voice search use cases where acoustic processing overperforms text-based search for audio files.

Social Media Listening

Today, people talk about brands on TikTok, Instagram, and YouTube videos more than they write about them. However, enterprises are still mainly focused on tracking written posts to protect their reputation and keep their competitive edge.

Speech-to-Text engines that struggle with proper nouns (e.g., brand mentions) cannot serve this use case adequately. Social media management platforms or brands can try to customize Speech-to-Text models to capture missing names and improve accuracy. However, after every customization, they may need to re-transcribe everything in the archive, which makes this model expensive and not sustainable.

Media and Entertainment

Audio and video content platforms such as streaming services, podcasts, or audiobook publishing services rely on limited text-based search based on the description instead of the rich audio and voice content they offer. For example, a user may want to find the show that has the quote “may the force be with you” or go to the moment Judy Garland says “there is no place like home” within The Wizard of Oz. These are only achievable with audio-based voice search.

Archiving

Some things are meant to be preserved in audio and video formats, such as memories from company events, family dinners or birthday celebrations. While converting these files into text is not needed, making them audio-searchable by indexing saves time and even memories.

If you’re interested in using Speech-to-Index for monitoring, discoverability, or archiving, work with experts at Picovoice Consulting. They’ll help you build the custom solution you need!

Consult an Expert