We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode EP 435: How 50X cheaper & faster AI transcription is changing enterprise work

EP 435: How 50X cheaper & faster AI transcription is changing enterprise work

2025/1/8
logo of podcast Everyday AI Podcast – An AI and ChatGPT Podcast

Everyday AI Podcast – An AI and ChatGPT Podcast
AI Deep Dive AI Chapters Transcript
People
J
Jordan Wilson
一位经验丰富的数字策略专家和《Everyday AI》播客的主持人，专注于帮助普通人通过 AI 提升职业生涯。
P
Philip Kiely
Topics
Jordan Wilson: 我认为我们没有充分讨论的一点是，我们所说的每一个词，我们进行的每一次对话，都非常有价值，而围绕这些对话的AI技术正在变得越来越便宜、快速和准确，这为各种规模的企业解锁了巨大的潜力。 语音转录可以将语音数据转化为文本数据，方便人们和机器进行处理，从而提高效率。 更便宜、更快的AI转录技术为企业提供了许多新的应用场景，例如客户服务、内容审核和媒体字幕生成。 尽管AI转录技术已经很准确，但在一些需要100%准确性的场景中，仍然需要人工进行验证。 Philip Kiely: Base10是一个AI基础设施平台，帮助客户部署各种AI模型，并优化模型性能，使其更快、更便宜、更高效。我们最近发布了世界上最快、最准确、最便宜的Whisper推理服务。 Whisper是由OpenAI开发的一个开源语音转录模型，具有高准确率和多语言支持。Whisper模型的最新版本速度更快，成本更低，可以实现实时转录。 AI转录的成本已大幅降低，从每小时1-2美元下降到几美分。可以通过将多个廉价的AI模型串联起来，构建更复杂、更经济高效的AI应用，例如AI电话接听。 更便宜、更快的AI转录技术正在推动可穿戴设备的发展，使长时间的语音记录成为可能。设备端推理和云端推理的性能差异导致了语音识别准确性的差异。 AI转录技术正在快速发展，未来会变得更快、更便宜、更准确，企业应该及早关注并尝试应用这项技术。

Deep Dive

Chapters

Shownotes Transcript

Send Everyday AI and Jordan a text message)

Meetings. Speeches. Quick thoughts to self. Those words are more than words. That's your company's secret sauce. Philip Kiely, Head of Developer Relations at Baseten, joins us to discuss.Newsletter: Sign up for our free daily newsletter)**More on this Episode: **Episode Page)**Join the discussion: **Ask Jordan and Philip questions on AI transcription)Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup)Website: YourEverydayAI.com)Email The Show: [email protected])**Connect with Jordan on **LinkedIn)**Topics Covered in This Episode:1. AI Transcription Benefits2. Whisper Model by OpenAI3. Cost of Transcription4. Business Applications for AI TranscriptionTimestamps:**00:00 Conversations are gold; AI makes them valuable.03:56 NVIDIA advances exceed Moore's Law; Apple's AI inaccurate.09:48 Text transcription technology error-prone; manual transcription necessary.11:19 Whisper V3: Low error rate, multilingual accuracy.14:58 Whisper rapidly transcribes audio with high efficiency.17:26 Emotion inflection crucial for text-to-speech synthesis.23:58 AI transcriptions need human verification for accuracy.25:35 Chain cheap AI models for efficient calls.30:53 On-device AI less powerful than cloud AI.33:07 Build prototypes now; technology improving rapidly.**Keywords:**Whisper by OpenAI, Automatic Speech Recognition, Open-source ASR, Accuracy, Multilingual ASR, MIT licensed, Amazon Transcribe, Whisper V3 Turbo, Live transcription, Speech inflection, ChatGPT, Philip Kiely, Jordan Wilson, Everyday AI podcast, Unstructured data, Anthropic funding, NVIDIA AI advancements, Apple AI alerts, AI transcription, Base 10, Searchable data, AI infrastructure platform, AI cost efficiency, Wearable technology, Voice control, On-device inference, Cloud inference, Speech synthesis, Business applications of transcription, Future of work

Learn how work is changing on WorkLab), available wherever you get your podcasts.