The success rate depends on factors such as audio quality, echoes, noise or background music, type of speech, register, whether it is dealing with standard language or a variant, volume, speed… Under optimal conditions, the success rate can exceed 95 % (see demos).
Live transcription is more complicated, so sometimes the results can be worse. The best idea is to test it in specific conditions. If it is not absolutely necessary to do it simultaneously, then it is better to record the audio and then transcribe it.
It accepts the most usual audio and video formats, such as WAV, MP3, MP4, MOV, WEBM, etc. You can also record from the website using the microphone on your computer or phone, without having to prepare any file.
A subtitle file (i.e., the transcription and the time stamps for each sentence or phrase, so that conventional audio or video players or web browsers can understand and display them) in different formats: SRT, VTT, etc.
Transcription with time stamps for each word (for example, for advanced searches in video).
Numbers, times, dates, abbreviations, acronyms, capitals, lowercase and punctuation signs ... will be correctly transcribed in all of them.
Yes, with nuestra API you can integrate Aditu into your workflow, application, CMS, etc. We also have an API for live transcriptions, which you can integrate into your virtual assistant, use to subtitle live events, etc.
Aditu, Elhuyar's speech recognition system, comes packed with new features both in terms of form and substance.To start with, we have changed its image.And now you can use it for bilingual transcriptions transcripts and transcription translations.