
What I value most about Google Cloud Speech-to-Text is its accuracy in real meetings, especially when several people participate and technical terms from marketing, campaigns, Google Ads, or ecommerce are mixed in. In our agency, we have tested it to transcribe meetings with clients, and it saves us a lot of time when reviewing agreements, pending tasks, and decisions made during the call.
I also like that it integrates well with other Google Cloud tools and that the processing is quite fast even in long meetings. It's not perfect, and sometimes you have to correct proper names or some specific words, but in general, the transcription comes out clean enough to work directly on it without having to listen to the entire meeting again. Review collected by and hosted on G2.com.
What I like least is that the quality of the transcription depends heavily on the input audio. When the client connects with a bad microphone, there is background noise, or several people speak at once, errors appear and then you have to spend a few minutes reviewing the text.
I also miss the automatic identification of different speakers being more consistent in dynamic meetings. In orderly conversations, it works well, but when there are interruptions or quick changes of turn, it can be difficult to follow who said what. Additionally, the system has many configuration options and initially requires some time to understand which parameters are most suitable for each type of meeting. Review collected by and hosted on G2.com.