Run a translated classroom

A lecture hall with international students presents a choice: teach in the academic working language (typically English) and accept that a portion of the room follows imperfectly, or segment students by language and duplicate instruction. Neither is satisfactory. Loquira introduces a third option — lecture in one language, let each student follow in theirs. For the broader picture on this approach, see Multilingual classrooms.

This guide covers the classroom-specific adjustments: microphone placement for a mobile lecturer, join code distribution without losing momentum, handling Q&A when the speaker rotates, and using transcripts as revision material.

Microphone placement for the lecturer

University lecturers move. They walk to the board, gesture at slides, step into the aisle to answer a question. The microphone must move with them.

Headset microphone is the correct choice for a lecture format (see Choose the right microphone for model recommendations). It stays at the corner of the mouth regardless of head movement, maintains consistent gain, and rejects the ambient noise of a 200-seat lecture hall. A lavalier is acceptable if the lecturer stays near the podium, but the moment they turn to write on the board the lavalier’s off-axis response degrades recognition.

Room microphone systems — ceiling mics, gooseneck mics on the lectern — are not sufficient. The recognition engine needs a speech signal that dominates the ambient room noise by at least 15 dB. Ceiling mics in a lecture hall typically achieve 3–6 dB above ambient at the speaker’s position. The engine will miss words.

Recommended setup:

• Lecturer wears a wireless headset microphone (Sennheiser ME 3-II or equivalent).
• The receiver connects to the lecturer’s laptop via USB audio interface.
• Loquira presenter view runs in the browser, capturing the interface input.

Test this setup in the actual lecture hall before the first session. The hall’s acoustics — hard surfaces, high ceilings, rear projection — affect the wireless signal and the recognition accuracy in ways that are hard to predict from a desktop test.

Sharing the join code

Lecturers have approximately 15 seconds at the start of class before attention fragments. The join flow must fit within that window.

Display the QR on the first slide of the lecture deck — not the title slide, the slide that appears while students are settling in. A slide that says “Open your phone, scan this code, select your language” with the QR centred at a legible size (at least 8 cm on the projected image). Add the session code in large type below the QR for students whose phone cameras struggle with the projection.

The join code can also be embedded in the LMS (Moodle, Canvas, Blackboard) as a persistent link for the course. Students who join from the LMS avoid the scanning step entirely. The link is static for the session duration.

Handling Q&A

Q&A is the hardest segment for any speech recognition system because the speaker changes unpredictably. A student asks a question from the tenth row, their voice reaches the lecturer’s microphone at a low volume, and the engine may not capture the question accurately.

Two strategies:

1. Repeat the question. The lecturer, after hearing the question, repeats it into the microphone before answering. This is the most reliable pattern — the engine captures the repetition cleanly, and the translation reflects the full interaction. Brief students that this will happen so they do not feel the lecturer is patronising them by repeating their words. For more on speaking clearly for the recognition engine, see Tips for non-native speakers.

2. Pass a handheld microphone. For small seminars (up to 30 students), passing a wireless handheld mic during Q&A ensures the engine captures the question directly. The lecturer retrieves the mic after the answer and continues. This pattern works well but requires a student willing to hold the microphone while speaking.

Strategy 1 is preferred for large lectures. It adds approximately 5 seconds to each Q&A exchange but guarantees transcript quality. Strategy 2 is preferred for seminars where the pedagogical style is conversational.

Transcripts as revision material

The session transcript, exported in plain text or JSON (see Curate transcripts after the event for cleanup best practices), becomes revision material for all students — not just those who followed in a translated language. A student who attended the lecture in English and wants to review a specific concept can search the transcript. A student who followed in translated French can compare the translation against the English original side by side.

For courses where attendance is graded, the transcript timestamp data provides a record of when each student joined and left the session. This is available in the session dashboard and can be exported alongside the transcript.

Recommended workflow:

• At the end of each lecture, the lecturer (or teaching assistant) ends the session and downloads the ZIP archive.
• Upload the transcript to the LMS alongside the lecture slides.
• For courses with weekly lectures, create a dedicated section in the LMS for transcript archives. Students build a searchable library over the semester.

Students who miss a lecture can read the transcript in their preferred language. This is not a substitute for attending — the transcript does not capture visual demonstrations, board work, or slide animations — but it closes the comprehension gap for students who would otherwise rely on a classmate’s notes.