AssemblyAI

AssemblyAI offers a range of AI models that can transcribe speech into text and extract insights from voice data. Models include speech-to-text, speaker detection, sentiment analysis, chapter detection, and PII redaction. The models are trained by AI researchers and are very accurate, especially the Universal-1 model, which has been trained on 12.5 million hours of multilingual audio data.

The service includes tools for developers to easily incorporate speech AI into their products. You can use it in a variety of ways, including a free tier for prototyping and pay-as-you-go for production. Pricing varies depending on the use case, but there are discounts for large-scale usage.

AssemblyAI's Speech AI models include:

• Speech-to-Text: Transcribe audio and video files with high accuracy.
• Streaming Speech-to-Text: Transcribe live audio streams synchronously with low latency.
• Speech Understanding: Analyze voice data and extract insights using audio intelligence models and large language models like LeMUR.
• Speaker Diarization: Identify the number of speakers in an audio file.
• International Language Support: Supports over 99 languages.

The service is geared for companies that want to create next-generation AI products that incorporate voice data. AssemblyAI offers 24/7 customer support and updates its AI models with the latest research and advancements.

Pricing tiers include:

• Free: For developers, includes transcribe up to 100 hours of audio, speaker diarization, custom spelling and vocabulary, and developer support.
• Pay as you go: Starting at $0.12 per hour for Speech-to-Text, includes unlimited access to Speech-to-Text, Audio Intelligence, and LeMUR, with concurrency starting at 200 files and 100 streams.
• Custom: Personalized plans with volume discounts up to 50%, solution architect support, and higher rate limits.

AssemblyAI takes security and privacy seriously, following standards like GDPR, PCI-DSS, and SOC 2 Type 1/Type 2, so you can trust your data is in good hands.

Published on June 14, 2024