Frequently asked questions

The interpreter's. In U.S. practice the reporter takes down the English interpretation and nothing else — official court-reporting guidance is explicit that the foreign language is not transcribed, that it is marked only as the witness speaking in a foreign language, and that correcting an interpretation is not the reporter's function. Most federal circuits will still treat that English as the witness's own statement under the language-conduit rule, but the Eleventh Circuit has rejected that rule and the others apply a multi-factor test that asks who supplied the interpreter and how qualified they were. The attribution holds until someone contests it. What we do is transcribe the French sitting in the audio, so that if it is ever contested you are answering with a document rather than an argument.

Yes, and no interpreter is needed for it. Under section 133 of the Constitution Act, 1867 either official language may be used before Quebec's courts, and each participant chooses independently — counsel may put the question in English and the witness may answer in French in the same breath. Both are the record. The transcript reproduces each language as it was spoken, with no translation, because nothing is being translated. This is the opposite of the U.S. model, and it is the situation we are built for.

Yes — this is the specialization. Counsel examines in English, the witness answers in French and crosses back mid-answer; we transcribe each language verbatim as spoken, without translating or paraphrasing, with the switch marked where the convention of the forum calls for it. Quebec French register, courthouse usage and legal terminology are handled by a practitioner who works in the law in both languages, not by a glossary applied afterwards.

Yes. Where an accused is tried in the official language of their choice under section 530 of the Criminal Code, section 530.1(g) requires the record of the preliminary inquiry or trial to include a transcript of everything that was said in the official language in which it was said, a transcript of any interpretation of it into the other official language, and documentary evidence in the language in which it was tendered. A dual-language verbatim record is not a preference in that setting; it is what the statute asks for. The same principle runs wider: a transcript that silently converts a proceeding into one language can undercut the very right being asserted. We produce records that keep the two languages distinct and attributable.

Order the audio or video and have the original-language answers transcribed and line-matched against the official English transcript, so the divergences are documented rather than asserted. That record supports errata, a motion to correct, or cross-examination on the point. Move quickly: U.S. courts have penalised parties who raised an interpretation problem late, in some cases treating the objection as waived, so this belongs on delivery of the transcript and not on the eve of trial.

Transcripts of recordings are a well-established category of evidence in their own right. For foreign-language recordings, U.S. federal model jury instructions treat the English transcript as the evidence the jury works from where accuracy is not in issue; where accuracy is disputed, the jury decides between competing transcripts by weighing the qualifications of whoever prepared each one, the clarity of the recording, and how the transcript was produced. That is why ours are produced and signed by a named practitioner with published credentials, why every transcript ships with a conventions page, and why uncertain passages are marked and timestamped instead of smoothed over.

Every Lexoral transcript carries a signed accuracy declaration attesting that the transcript is a faithful and complete account of the audio provided, signed by the practitioner who produced it. Forums differ on the form of attestation they expect, and the form that satisfies a Quebec filing is not the one a U.S. filing expects. Tell us where the transcript is going at intake and we will confirm what your filing needs before you engage us.

Because when a translation is disputed the court does not weigh the vendor — it weighs the person who made it. Model instructions direct jurors to consider the qualifications of the translator, the clarity of the recording and how the transcript was produced. A transcript from an anonymous pool has nothing to put in that column. Ours is produced by a practitioner who has delivered legal argument in French before Quebec's National Assembly and written constitutional-law texts in both languages, and who can say so under oath.

Yes. Continuous line numbering, structured pagination and speaker identification, formatted for direct reference in pleadings and briefs — so a citation made once holds for the life of the matter.

Tell us the page and line. The passage is re-audited against the source recording. Where the audio does not support what was transcribed, it is corrected and the declaration re-issued at no charge. Where the audio does support it, we tell you that plainly and mark the passage as disputed rather than quietly changing the record — because a transcript that moves on request is not evidence.

By convention, visibly, and never by guessing. Overlapping speech is marked as such (voix superposées / simultaneous speakers) rather than invented into clean sentences. Inaudible passages carry a timestamp — [inaudible 00:14:22] — so you can go to the audio yourself. Uncertain proper nouns are marked [phonetic]. Silences of evidentiary length are timed. Every transcript ships with a conventions page so opposing counsel reads the same marks the same way.

Exhibits are carried into the transcript as they are marked on the record, with the announcing speaker attributed. Where a cross-border matter runs Quebec pièce numbering on one side and U.S. Exhibit numbering on the other, both are carried in the same record so neither set of counsel has to translate references by hand.

AI transcription fails precisely in the conditions that define bilingual legal proceedings: mid-sentence code-switching, overlapping speakers, legal terminology, proper nouns and Quebec French register. It also invents fluent text where the audio is unclear, which is the single worst failure mode in an evidentiary record. More fundamentally, a model cannot maintain a chain of custody, cannot be cross-examined, and cannot sign an accuracy declaration.

Yes. Emergency-services audio, custodial and police-station interviews, and body-worn recordings are regular work, including for public-defence offices. This is difficult audio — crosstalk, radio traffic, distress, background noise — and it is quoted after we have heard it, not from a duration estimate.

Two named practitioners — language and verbatim discipline on one side, Quebec procedure and jurisdiction on the other. Files are not distributed to a contractor pool or an offshore desk, and nothing is delivered as reviewed that was not in fact reviewed by a person. The name on the declaration is the person who did the work.

Five business days as standard, forty-eight hours on priority. Multi-day proceedings are scheduled individually, and where a hearing runs over consecutive days we can deliver day-by-day so counsel is not waiting on the full record to prepare the next morning.

Per page, quoted in Canadian dollars, after we have heard the recording. Length alone does not set the price — overlapping speech, poor audio and sustained code-switching change the work considerably, so a firm figure follows the audio review rather than preceding it. Confirmation of receipt is the same working day.

Searchable PDF and editable Word, line-numbered and paginated, with speaker identification, a conventions page and the signed accuracy declaration. Delivery is by encrypted transfer. Formatting corrections are made at no charge.

Most audio and video formats, including MP3, WAV, M4A and MP4. If a recording is difficult in places we say so before quoting, not after delivering, and we identify which passages carry risk so you know where the record is soft.

Yes. Files are handled through restricted-access workflows in compliance with Loi 25, transferred by encrypted protocols, stored in restricted-access environments, and never disclosed to third parties except where you instruct us to involve one. Clients may request permanent deletion of recordings and transcripts at any time.

No. Client materials are used exclusively for the transcription assignment. They are never shared, repurposed, sold, or used to train or develop any model or system, internal or external.