Streaming speech: protect personal data before transcription
Real-time ASR and translation models can process speech containing personal data; Rota Nacional’s privacy barrier applies the organization’s policy before a model runs.
The announcement about Youdao’s ASR and translation models highlights streaming and the option to chain transcription with translation. In live speech applications, however, audio and transcripts may contain names, phone numbers, and other personal data.
At Rota Nacional, audio transcription jobs pass through the privacy barrier before a model runs: depending on the organization’s policy, detected personal data can be replaced with markers, removed, or blocked. The platform offers an API compatible with OpenAI and Anthropic; compatibility with the models mentioned in the announcement is not implied.