# How to transcribe a conversation, step by step > To transcribe a conversation, ask permission before recording, capture clear audio, and check that the file plays correctly. Upload it to Kotosay, let the language be detected or select it, then compare the text with the recording. Correct important details, choose a word-for-word or cleaned-up version, and copy or export it. Learning how to transcribe a conversation starts before you upload a file. A recording with clear voices gives you a better starting point, but the useful work continues when you check the text against what was actually said. Names, dates, unfinished sentences, and small words such as “not” deserve particular attention. For a class assignment, a research conversation, or a reported interview, the aim is the same: produce text you can return to without losing the meaning of the recording. ## How to transcribe a conversation step by step Follow these steps from preparing the recording to saving the checked text. If you already have a recording, start by checking the file, then work through the upload and review stages. ### 1. Set the purpose before you record Ask before you record, because recording rules differ by country and state. Decide what you need the finished text to do. A transcript for close language analysis needs different treatment from a readable conversation published on a blog. A student may need an accurate record for an assignment, while a journalist may use the text to locate and verify potential quotes. Write down any requirements you have been given. These might include preserving filler words, noting meaningful pauses, or keeping the wording unchanged. If you are working from a research protocol or a publication's style guide, use that as your reference throughout the review. Keep a separate note of expected names, organizations, and specialist terms. That list will help you check spellings later, but it should not override what you hear in the recording. ### 2. Record close enough to capture clear voices Place your phone or recorder close to the people talking, rather than at the far end of a table. Keep the microphone unobstructed and avoid putting the device beside a laptop fan, air conditioner, or open window. Choose a quiet room when you can. Background music, dishes, traffic, and echo can make words harder to hear. A soft-furnished room is usually a better choice than a large, bare space. Make a short test recording and play it back through headphones. Listen for muffled words, distortion, and distracting background sound. Adjust the position or input level before starting the full conversation. During the recording, avoid handling the device or placing papers over it. For an online conversation, wear headphones to help prevent sound from the computer being picked up again by the microphone. Clear source audio reduces guesswork during review. ### 3. Check the file before uploading it Open the saved recording and check the beginning, a section in the middle, and the ending. Confirm that it contains the conversation you expected and has not stopped partway through. Then check its duration and file extension. Kotosay accepts audio files in MP3, M4A, WAV, AAC, OGG, OPUS, FLAC, AMR, WMA, and AIFF formats. It also accepts MP4, MOV, WEBM, MKV, and AVI video files, so a supported video recording does not need its audio extracted first. If your recorder produces an unusual format, use the free [audio converter](https://kotosay.com/audio-converter) to convert it to MP3 if the converter can open it. The conversion runs in your browser. Play the converted file before uploading it to make sure the sound is intact. Keep the original recording as your reference. Converting a file can make it compatible, but it cannot restore words that were never captured clearly. ### 4. Upload the recording and select the language Open Kotosay's [page for turning a recorded conversation into text](https://kotosay.com/audio-to-text) and upload your supported audio or video file. Kotosay is a website, so there is nothing to install. You can let the language be detected automatically or select it yourself. If you know the recording's language, choosing it gives you an explicit setting to check before proceeding. Check the file length against the access available to you. Without an account, the first file shows its first 5 minutes. Sign up for free to read the rest, up to 30 minutes. A free account requires no card and allows 3 files a day, processed one at a time, with up to 30 minutes per file. For a longer recording, the Unlimited plan supports files up to 10 hours. Choose based on the actual file you have rather than assuming a long conversation will fit the free limit. ### 5. Read the text against the recording Treat the generated text as a draft to verify, not as a finished record. Start at the beginning and read while listening to the recording. Working in order helps you keep the context around a sentence instead of correcting isolated words. Use the timestamps to return to passages that need attention. When something sounds wrong, replay enough of the surrounding audio to understand the sentence. A word that seems plausible on its own may not fit the question or response around it. Pay particular attention to quiet passages, interruptions, and sentences spoken over background noise. Do not fill a gap with what you think the person probably meant. If a passage remains unclear, use a consistent note such as “[unclear]” rather than presenting a guess as certain. Complete this listening pass before polishing the prose. Otherwise, you can spend time improving a sentence that does not match the recording. ### 6. Correct names, numbers, and technical terms Use the editor on the page to fix the wording. Kotosay's editor is free for everyone, and the timestamps stay when you correct words. Check proper names against a reliable reference, such as an organization's own website or information supplied for the conversation. Listen again before replacing a similar-sounding word with a name from your notes. Context helps, but the recording remains your source. Give numbers their own review. Dates, amounts, measurements, and percentages can look perfectly reasonable even when they are wrong. Check units as well as values, and listen for qualifiers such as “about,” “at least,” or “less than.” Technical vocabulary deserves the same care. A specialist term may have been rendered as an everyday phrase. Use your preparation notes to confirm the spelling after checking the audio. Finally, look for small changes that reverse meaning. “Can” and “can't,” or a missing “not,” can matter more than an awkward paragraph break. **Correct meaning before polishing style.** ### 7. Choose a word-for-word or cleaned-up version A word-for-word transcript keeps the spoken wording, including repetitions, filler words, and false starts. This can be appropriate when the way someone speaks is part of what you are studying, or when an assignment requires a close record. A cleaned-up transcript removes verbal clutter so the conversation is easier to read. You might remove an abandoned start or a repeated “you know,” while leaving the sentence's meaning and tone intact. For example, “I, um, I thought the meeting was on Tuesday” can become “I thought the meeting was on Tuesday.” But “I think it was Tuesday” should not become “It was Tuesday.” The uncertainty belongs in the text. Apply your choice consistently. Do not keep every hesitation in one passage and heavily rewrite the next. If you need both versions, retain a checked word-for-word copy before making readability edits. For research work, follow the project's requirements. For direct quotes, verify the selected wording against the recording before using it elsewhere. ### 8. Copy or export the checked transcript Once the wording is checked, copy the text into your working document or download it with the Unlimited plan. Unlimited includes downloads as TXT, DOCX, PDF, SRT, and VTT. Choose a format that suits the next task. TXT gives you plain text, DOCX suits further document editing, and PDF works for sharing a fixed-layout document. SRT and VTT are subtitle formats, rather than the usual choice for a written conversation. Name the file so you can distinguish the checked transcript from an earlier draft. If you made a cleaned-up version, make that clear in the filename or document heading. Open the saved copy and check that the text you need is there. Keep enough context around any passage you move into notes, an assignment, or a story so its meaning remains clear. ## A short checklist before you use the text - The recording plays correctly from beginning to end. - You have read the transcript while listening to the audio. - Names, numbers, technical terms, and negatives have been checked. - Unclear passages are noted rather than guessed. - Word-for-word or cleaned-up editing has been applied consistently. - The saved version is clearly named and ready for its intended use. Keep file retention in mind, too. Kotosay automatically deletes the audio file 24 hours after transcription, while the transcript stays until you delete it. Keep your own recording if you will need to verify a passage later. A checked transcript is useful for finding material, but the original audio is still your reference for what was said. ## FAQ ### Should I keep filler words in a conversation transcript? Keep them when your assignment, research method, or intended use requires a word-for-word record. For a readable version, you can remove fillers that add no meaning. Preserve hesitation or uncertainty when it affects the message, and apply the same editing approach throughout the document. ### Can I transcribe a conversation recorded on my phone? Yes, if you can upload the saved recording in a supported format. Kotosay accepts common phone recording formats such as M4A, MP3, and WAV. Check that the file plays correctly before uploading, and keep the original recording so you can verify unclear wording during your review. ### What should I do when I cannot understand a word? Replay the passage with the surrounding sentence, then check any relevant names or technical vocabulary in your notes. If you still cannot hear it confidently, mark it as unclear rather than guessing. Keep the nearby timestamp so you can return to that point in the recording. ### Can I check and edit the transcript for free? Yes. Kotosay's on-page editor is free for everyone, and timestamps stay when you fix words. Without an account, the first file shows its first 5 minutes. A free account lets you read up to 30 minutes per file, with 3 files a day, one at a time. ### How do I save the finished conversation as a document? Copy the checked text into your working document, or use the Unlimited plan to download TXT, DOCX, or PDF files. Choose DOCX for further editing or PDF for a fixed-layout copy. Open the saved file afterward to confirm that it contains the version you intended to keep. Noemi Castell, Interviews and podcasts editor 2026-11-11 · Kotosay · https://kotosay.com/blog/how-to-transcribe-a-conversation ## Other languages - [Português](https://kotosay.com/pt-br/blog/como-transcrever-uma-entrevista.md) - [Español](https://kotosay.com/es/blog/como-transcribir-una-entrevista.md) - [Bahasa Indonesia](https://kotosay.com/id/blog/cara-transkrip-wawancara.md) - [Français](https://kotosay.com/fr/blog/comment-transcrire-un-entretien.md) - [Italiano](https://kotosay.com/it/blog/come-trascrivere-un-intervista.md) - [Deutsch](https://kotosay.com/de/blog/gespraech-transkribieren-schritt-fuer-schritt.md) - [Türkçe](https://kotosay.com/tr/blog/roportaj-nasil-yaziya-dokulur.md) - [العربية](https://kotosay.com/ar/blog/how-to-transcribe-a-conversation.md) - [日本語](https://kotosay.com/ja/blog/how-to-transcribe-a-conversation.md)