May 2021 Summaries

Deepgram has become a founding member of CallMiner's Open Voice Transcription Standard (OVTS), which allows customers to choose the best speech recognition provider for their specific use cases. The OVTS API facilitates integration with any automatic speech recognition solution on the CallMiner Eureka platform, offering businesses more options in ASR providers based on their needs. This new standard follows a trend towards flexible, best-of-breed solutions tailored to individual business requirements and use cases. Deepgram's End to End AI Speech Platform can provide real-time transcriptions for conversational analytics and custom speech models for domain-specific accuracy.

In celebration of Asian American & Pacific Islander Heritage Month, six Asian American & Pacific Islander Deepgrammers share their personal stories and cultural values. The group includes individuals with Chinese, Vietnamese, Hong Kong, and Filipino heritage who discuss their experiences growing up in America, the importance of family bonds, and the challenges faced by immigrants. They also emphasize that being AAPI is not a monolithic experience but rather a diverse range of cultures and histories tied together by common themes such as perseverance, hard work, and respect for personal sacrifice. The conversations highlight the shared values among AAPI communities and their contributions to American society.

The text discusses the use of Python with speech-to-text technology, specifically focusing on Deepgram, a voice recognition company that transcribes audio into machine-readable data. It highlights how unstructured data like voice and video can be transformed into valuable insights using this technology. Deepgram's innovative architecture uses End-to-End Deep Learning Speech Recognition to provide accurate transcripts and improve over time. The company offers various features such as real-time or streaming transcription, recorded or batch audio transcription, noise reduction, and diarization. Deepgram also has a Python SDK available on Github for developers to get started with the technology.

Deepgram has released new Spanish and Turkish language models along with updated general models. The company's Production Team and Deepgram Labs are responsible for developing these models, which can be tailored to specific use cases. Additionally, the Conversational AI Model is a limited release model that can be trained for individual needs. As more customer audio data becomes available, these lab models will transition to general release status. Interested users should contact Deepgram for access. Feedback on this post or any other aspect of Deepgram's services can be provided through GitHub discussions.

"Automatic Speech Recognition (ASR): Past, Present, and Future" is an ebook that delves into the world of ASR, a powerful speech to text solution. The book explains what ASR is, its importance in extracting value from unstructured data, and how it works through three types: Traditional Method, Hybrid Method, and End-to-End Deep Learning Method. It also explores the future of ASR and how organizations can prepare for a Unified ASR Solution. The ebook aims to provide insights into the potential benefits and applications of ASR in various industries.