Latest from Talking Owl · ReadAlongs
Digitizing an Indigenous language recording protects its signal, but preservation is not the same as use. When a community wants a recording to support learning, a ReadAlong can bring an authorized audio file and reviewed transcription together: the words highlight as the voice plays, and learners can select a word to hear it again.
The experience is straightforward for a learner, but creating a resource that is accurate, appropriate and maintainable takes careful decisions around purpose, authority, orthography, timing, access and handover. The National Research Council Canada describes ReadAlong Studio as an open-source tool developed with Carleton University and Indigenous collaborators for aligning existing audio or video with a transcription.
Begin with purpose and authority
Before choosing a recording, decide who the resource is for and where it may be used: a classroom, a community network, a restricted local system or the public web. Identify who can approve the recording, transcription and intended access, and keep unresolved restrictions visible. Permission to record or digitize does not automatically authorize transcription, alignment, publication, AI training or another new use; a ReadAlong is a derivative that needs its own review.
These decisions should remain connected to the people responsible for the language and collection. Talking Owl’s data sovereignty and community-control guide offers questions for documenting purpose, permission, custody, access and handover before files are sent to a tool or made available to learners.
Prepare the recording and the words together
A useful pilot begins with one recording that the team has authority to work with and one transcription reviewed by people who know the language. Consistent Unicode text, agreed spelling conventions and clear file identifiers make technical work easier, but alignment should not be treated as a way to correct the language. The recording and written text carry their own histories, and differences between them may need linguistic or community interpretation rather than an automatic fix.
Treat automatic alignment as a draft
ReadAlong Studio uses grapheme-to-phoneme conversion to represent how written forms relate to speech sounds, then matches that information against the recording to estimate timing. The result can save substantial manual placement, but it still needs review by speakers, educators or other people authorized to assess the language and its presentation.
My 2024 MA work developed Kʷu sqilxʷ: an interactive digital edition, bringing nsyilxcən recorded speech and written text together through ReadAlongs. That work, and related collaboration on grapheme-to-phoneme mappings, continues to shape the practical support I offer for preparing files, checking alignments and building a workflow a language team can maintain. The public Research & ReadAlongs page links to the thesis, a published paper and an authorized example.
Design the handover before scaling up
A finished pilot should include the source files, the reviewed transcript version, the alignment output, a record of corrections, access decisions and clear instructions for making the next resource. Agree who will maintain the files, where they will be stored and how future changes will be reviewed. Training matters here: the strongest outcome is not a single polished demonstration, but a process the community can understand, revise and continue under its own direction.
A practical first step
Choose one short, authorized recording and a reviewed transcript. Agree where and how the result may be used, build a draft ReadAlong, review its words and timing with the right people, and document every correction. Scale up only after the team is comfortable with the resource and its handover.
If your team is considering a ReadAlong project, Talking Owl can help plan the pilot, prepare the materials and develop a repeatable training workflow. Explore consulting and capacity building, or begin with a free introductory conversation.
Leave a Reply