Only institution-level Administrators have access to this ASR option.
This feature was formerly a toggle labeled "Automatic Language Detection" but has been changed to allow for specific language selection if desired.
Echo360 has partnered with several transcription service companies to provide an automatic speech recognition (ASR) service. The ASR service sends media to the transcription provider to automatically create machine-generated transcriptions for video/audio media. Once transcribed, the media can be played with the transcription panel open, allowing users to both see and hear the text of the speech in the video.
This article discusses the ASR Language Settings available in Echo360. It allows you to select the primary language for your transcriptions. Meaning if you know all (or most) of your institution's recordings will be in a specific language, you can identify that language as the one to use for transcriptions.
Some institutions may have the option to enable Automatic Language Detection. This is discussed in more details in the section at the bottom of this page.
The language selected by default is English (US, UK, or AU/ANZ region-specific as appropriate). Administrators configure this setting at the institution level, meaning that it applies to all media sent for transcriptions.
Important: This is not a translation setting; this simply identifies the language typically being used in the recordings. For example, French spoken in a primarily English language class will look like gibberish in an English language transcript. If you select French as the transcript language setting, the French will be transcribed accurately. Any OTHER language spoken will be gibberish in the transcript. These are machines and do not make language adjustments like humans do.
Providing language-specific transcriptions for media allows you to more easily apply translated closed captions to media, by reviewing and editing the transcript then applying it as closed captions. Furthermore, users can always manually edit the transcriptions to clean up the non-translated segments of an automated transcript.
To configure language settings for automated transcripts
- Log into Echo360 as an Institution-level administrator.
- Click the Settings icon and select Institution Settings.
- Select Features from the options in the right panel, as shown in the below figure.
- Scroll down to the Transcripts and Captions section of the page.
- Click the ASR Language Settings button, identified in the above figure.
The dialog box that appears contains a Default Language Selection selection drop-down list. It may also contain an Automatic Language Detection toggle, shown in the below figure and discussed later on this page.
- Use the Default Language Selection drop-down list to select the language to be used for all of your institution's media transcriptions.
- If you have the option, you can enable the Automatic Language Detection toggle.
If enabled, you do not need to select a Default Language from the drop-down list. This is discussed below.
- When finished, click Save.
See Automatic Transcriptioning Service or ASR for additional details surrounding the transcription service. See Using the Echo360 Transcript Editor for more information and documentation to provide users for editing transcripts.
Using Automatic Language Detection
Some institutions will have an Automatic Language Detection toggle in the ASR Language Settings modal. This is shown in the figure in the above procedure.
If turned ON, Echo360 tells the transcription service to automatically detect the PRIMARY language spoken in the audio of the media, and provide the transcript in that language. You do not need to select a default language if you enable automatic language detection. Though it does no harm if you do.
Automatic language detection provides transcripts only in languages supported by Transcribe, which are listed in the following article: https://docs.aws.amazon.com/transcribe/latest/dg/what-is-transcribe.html.
If the language you want is NOT supported by Transcribe, do not enable this toggle, and instead select the language from the Default Language Selection list.
It is important to note that auto-language detection for transcriptions is an all-or-nothing prospect; each transcript will be provided in ONE and ONLY ONE language (the language used in the majority of the speech). If the media is 70% Spanish and 30% English, the transcript will be provided in Spanish, with the Spanish speech transcribed as accurately as possible. The English portions will be mostly gibberish, with low-confidence (machine-driven) guesses as to "what Spanish word is being spoken here?" This feature is NOT designed for "mixed-language" classes and may not provide useful transcripts in those cases.