Converting tapes and other fagile media to digital audio files is almost a trival matter today. But what format should be used for long term storage? What metadata should accompany the files? How can or should they be accessed?
Thinking about using compressed prorietary formats or a reduced interchange metadata format? Wondering how much work will it intail to create an open standards based audio archive?
The ResCarta Toolkit can help build a true audio archive. With Broadcast Wav format, Library of Congress audioMD and MODS metadata, checksum validation over time and an almost instant website to allow for discovery and access; the ResCarta Toolkit is your answer to organize your audio history.
Automatic Audio Transcription (AAT) is contolled by a single check box on the Data Conversion Tool. Under the covers the ResCarta Toolkit uses the open source SPHINX software from Carnegie Mellon University of Pittsburgh. This software converts digital audio to text much like printed text is converted to characters and words by Optical Character Recognition software (OCR). The ResCarta Toolkit currently supports English as the language output and uses the Wall Street Journal dictionary to provide more recognition of family names.
Recent software developments by OpenAI have produced WHISPER tools for creating SRT files and TXT files from audio and video formats. Whisper supports many languages and is more accurate than our internal SPHINX. We here at the foundation have adapted the DATA CONVERSION TOOL to accept SRT files from WHISPER tools. To do so just place the SRT and the WAV file or SRT and MP4 file in a directory and UNCHECK the AAT check box. The DCT will read the SRT into the resulting ResCarta Object and make it searchable.
We also advise the use of another tool called SRTeditor from Niskee This tool includes WHISPER and is an excellant SRT tool.
The Audio Transcription Editor (ATE) allows the automatically produced transcription to be corrected for final inclusion in your digital archive. Each word is stored within the Broadcast Wave formatted file along with its timecode so that word search in your audio files can be accomplished.
Broadcast Wave format or BWF is a standard format that can be created without royalty payments and has the featured needed to add required metadata. It is the format used for radio and television production.
The ResCarta Toolkit can create BWF format from common WAV formatted files. The Data Conversion Tool will embed metadata made by the Metadata Creation tool into the aXML chunk according to the Broadcast Wave Specifications. Transcription data created by ResCarta Toolkit's unique one click Automatic Audio Transcription (AAT) is stored within the BWF file along with the location of each word similar to the way Optical Character Recognition (OCR) data is stored in textural materials.
Files created by the ResCarta Toolkit can be validated using the BWF MetaEdit from Library of Congress and audio visual working group of the Federal Agencies Digitization Guidelines Initiative (FADGI)
Audio resolutions are currently supported up to 96khz 24bit Stereo by the Data Conversion Tool.
The Metadata Creation Tool (MCT) uses simple forms to collection and output Library of Congress METS/MODS/audioMD formatted XML files for each of your original WAV formatted audio files. The Data Conversion Tool will take your WAV formatted file and create a standard directory structure where it will store the audio chunk into a Broadcast Wave formatted (BWF) file. It will take the MODS data created by the MCT and embed that data into the aXML chunk of the BWF. If the Automatic Audio Transcription (AAT) checkbox was used or the Audio Transcription Editor (ATE) during production of your ResCarta archive, then the transcribed text will also be embedded into the BWF file.
Oral historians have been transcribing audio since the creation of audio recording devices. Thinking about digitizing these and making them accessible? The ResCarta Toolkit can convert scanned paper documents into fully searchable digital objects. Have the transcription in a proprietary word processing format and have the original audio file, then use Audacity to convert your existing audio file to WAV format and add metadata using the Metadata Creation Tool. Open your word procssing file, copy the text and use a NOTE field to store the transcription. Name the NOTE field to "Transcription" and it can be fully indexed by the ResCarta Indexer and used as a search field in ResCarta-Web. Go even further by checking the "Enable Audio Transcription" box on the Data Conversion Tool and have your audio automatically transcribed (English only for now). The full transcription can be editied and corrected with the Audio Transcription Tool. Now your archive will contain a standard Broadcast Wave formatted file complete with your word processed transcription and the automatic transcription with each word and timestamp stored within.