In today's fast-paced business landscape, efficient and cost-effective solutions are paramount for success. Dubverse has high transcription loads from their clientele, who also expect fast precise transcripts. Dubverse needed to deploy efficient models in an optimized architecture for success.
The Challenge:
DubVerse, a leading media company, has more than 120k users. Most of these users seek to transcribe videos resulting in high transcription costs. This started affecting the unit economics for Dubverse. Scaling up to serve such a large number of users concurrently was also something that most legacy ASR services could not keep up with. Seeking a transformative solution, they turned to SimpliScribe for a more efficient and cost-effective alternative. Their transcription loads cost them ~$3000 a month, from a leading ASR tool. They were running it as an API call on their platform so they would re-route most transcription requests to the third-party service.
The Solution:
SimpliScribe's advanced transcription service proved to be a game-changer for Dubverse. By leveraging model-level optimizations and fine-tuning, SimpliScribe offered a custom deployment of the Whisper (OpenAI) model on the premises of Dubverse. The implementation was seamless, allowing them to experience the benefits from day one. On-premises implementation ensured data privacy which was crucial for their compliance requirements.
Cost-Effectiveness in Action:
SimpliScribe delivered cost savings that exceeded DubVerse's expectations. Compared to their previous transcription service, SimpliScribe's deployment resulted in a direct 10x reduction in costs. This substantial cost-effectiveness not only boosted Dubverse's bottom line. They ended up allocating resources to other critical aspects of their business while we focused on what we do best - deploy kick-ass ML models to empower products.
Unparalleled Speed and Accuracy:
Aside from cost savings, SimpliScribe's transcription solution also provided exceptional speed and accuracy. Using our optimized serving suite we were able to transcribe 36% faster on average, giving Dubverse’s users an accelerated video editing experience, while reducing compute costs. Our custom-trained transcription model captures phonetics rather than words making it more accurate, we deploy it on our proprietary serving suite which boosts performance through kernel mapping.
What this means for you-
While advances in AI have helped us make leaps in Speech Recognition technology, there is still a large room for savings and improvement in accuracy if you deploy your models right. If you deploy/want to deploy a transcription model to empower your product line-up, we can improve accuracy and reduce costs for you too. For more information visit - https://simpliscribe.simplismart.ai
