Latest news and updates from DolphinVoice.
FeaturedDolphinVoice focuses on the Japanese market, providing speech recognition, pronunciation assessment, API, SaaS, and smart device solutions for businesses, education, and multilingual service environments.
A “real-time” speech recognition API may return text during an utterance—or wait until the sentence ends. Compare pure and pseudo-streaming by result timing, connection model, and interim results.
DolphinSOE has released a Japanese pronunciation assessment API, providing personalized speaking assessment services for Japanese learners.
This article introduces the speed indicator in real-time speech recognition: Tail Packet Latency. DolphinVoice provides the best user experience for real-time speech recognition scenarios through extreme tail packet latency optimization.
This article will guide you on how to quantitatively evaluate the transcription speed of audio files and explore the role of parallel processing in improving the transcription speed of audio files.
When evaluating the performance of speech recognition systems, CER and WER are two very important metrics. This article introduces the definitions, calculation methods, and limitations of these two metrics, emphasizing the need to consider other indicators for a comprehensive assessment of speech recognition engine performance.