LONG RECORDING · MP3, WAV AND M4A

Convert long audio to text.
Keep the timestamps.

Upload a recorded MP3, WAV or M4A and get timestamped, editable text. The quick form accepts up to 90 MB on Free; eligible Premium uploads up to 1 GB use the main app's chunked uploader. Keep the tab open until upload finishes.

20 transcriptions + 60 min / month Try without an account Handles long recordings No install Source file kept temporarily, then deleted automatically

Choose an audio recording Free account: 60 min/month

Free account vs guest demo: Sign in for up to 20 recordings and 60 monthly minutes, subject to plan and per-recording limits. Guest processing may provide a shortened preview rather than a full recording. Full TXT, DOCX, SRT, VTT and JSON exports require a signed-in account and an eligible completed job. This direct upload accepts files up to 90 MB. Premium chunked upload supports up to 1 GB in the main app. File size and monthly minutes are separate limits.

This quick form accepts files up to 90 MB. Premium allows up to 1 GB through the main app's chunked uploader. Keep this tab open until the upload completes.

FROM RECORDING TO EDITABLE TEXT

Keep the timestamps. Lose the manual typing.

Review names and technical terms in the browser, search the recording, and download a clean transcript or timed subtitles.

TXTSRTVTTDOCXJSON

Example transcript

00:00Upload the recording directly from your computer.

00:14Long audio continues processing while you are away.

00:29Return to searchable text with timestamps and exports.

How do I convert an audio file to text?

Upload a supported MP3, M4A, WAV or other audio recording and review the transcript after processing. Download TXT, DOCX, JSON or separate SRT subtitles. Size, duration and plan limits apply.

HEAR IT · CHECK THE ACTUAL OUTPUT

Hear a recording, then read the actual transcript

This is a real, preserved primary-ASR result from our short licensed benchmark, not an invented customer quote. Listen and compare the spoken words with the actual output. Long multi-hour files have separate eligibility and size limits.

Listen to a real 10.44-second source recording

Source: LibriSpeech test-clean, CC BY 4.0. Recorded speech reading, not a client recording or live lecture.

Actual LongScribe primary-ASR output · Sept 11, 2026

00:00:00,000 → 00:00:10,040 He hoped there would be stew for dinner, turnips and carrots and bruised potatoes, and fat mutton pieces to be ladled out in thick, peppered, flour-fattened sauce.

Recorded with a CC BY 4.0 source from the frozen LibriSpeech benchmark and transcribed by LongScribe's Groq/Whisper primary-ASR path on September 11, 2026. The text shown here is the preserved model output, not invented marketing copy. This short clip is not evidence of long-file accuracy, or of subtitle timing quality or a different plan.

Which upload route works for your file?

File up to 90 MB

Choose a supported audio file in the form above. Signed-in Free has 20 recording jobs and 60 minutes per month, subject to admission rules.

Choose a file above

Larger eligible Premium file

The quick form cannot accept files above 90 MB. Premium has a separate chunked uploader for eligible files up to 1 GB and costs $4.99/month. This is not an unlimited upload promise.

Open larger-file uploader

How SRT subtitles work Supported public links Compare plans Open the main app

LONGSCRIBE PREMIUM

Need more than 20 audio recordings a month?

Free offers 60 minutes across 20 recordings per month. Premium is $4.99/month for up to 5,000 transcription minutes and 500 recordings per month. Separate per-job and file-size limits still apply.

$4.99/month

Explore Premium → Cancel any time

Official pricing sources

Official provider sources for verification. See the dated comparisons below; prices and limits can change.

Uploading audio in the browser

The file picker on this page starts an upload, not a streaming microphone session. Keep the page open while the upload is in progress. After LongScribe accepts the job, transcription continues in the background and you can return to your results.

Supported audio formats and public recording links

The file workflow supports common recorded audio types, including MP3, M4A, WAV, FLAC, AAC, OGG and Opus. To transcribe a supported public YouTube, SoundCloud or Vimeo recording, use the main app's URL importer instead of the file picker. Private, deleted, paywalled or inaccessible sources may fail; use an authorized original file when you have one.

Reviewing accuracy, accents and overlapping voices

Recognition quality depends on language, sound quality, background music, accents, proper names and speakers talking over one another. The AI result is a draft: listen to important passages and verify names, numbers and quotations before publication. Speaker labeling is available on eligible Premium plans.

Editing words does not retime the recording

You can correct recognized transcript text in the editor. The original timestamp boundaries remain: manual text editing does not change when a person spoke in the audio. Where a manual edit preserves the original segment alignment, timed SRT/JSON exports may retain those boundaries with corrected words.

TXT, Word, SRT and JSON exports

Choose TXT for plain notes, Word DOCX for a document, SRT for a separate timestamped subtitle file or JSON for structured downstream processing. An SRT file does not burn captions into an audio or video recording. The examples above are format illustrations, not a measured accuracy result.

Audio transcription for long recordings

Multi-hour audio files can enter LongScribe's background queue, but that does not bypass upload-size, per-job, account or monthly-minute limits. You must finish uploading before leaving the page. For a multi-hour project, check your eligible plan and actual source size first. For repeated files, compare the recurring Premium allowance before upgrading.

Upload size limits by plan

Direct uploads are capped at 90 MB on the Free plan; uploads can be up to 1 GB on Premium (including Premium Pro) via chunked upload. Supported public URLs are fetched server-side and do not use the browser direct-upload limit.

Recording permission and privacy

Only upload audio you have permission to transcribe. Uploaded source media is retained temporarily for processing and deleted under LongScribe's source retention policy. Private transcripts are not automatically published; sharing requires an explicit action. Check the product privacy policy before submitting confidential recordings.

Transcribing noisy audio and non-native English speech

For noisy or accented audio, first improve the recording if you can, transcribe a short representative excerpt, and review names, numbers and uncertain phrases against the source before exporting. In LongScribe's September 2026 standard-ASR benchmark v1, the noisy/far-field slice had 30.11% WER (28 errors / 93 reference words) and the non-native English slice had 19.78% WER (18/91). These are self-published September 11, 2026 results from the then production-primary Groq Whisper large-v3-turbo ASR path on only five short speech clips per slice. They do not measure a whole long file, current accuracy on every language or speaker attribution.

Read this audio workflow in another language

Frequently asked questions

How do I convert an audio recording to text?

Choose an eligible local audio file in the form, wait for the upload to finish, review the generated transcript and export TXT, DOCX, SRT or JSON. The accepted transcription can run in the background.

Is audio to text the same as speech to text?

For an existing recording, yes: both describe automated audio transcription of spoken words. LongScribe is for saved files or supported public recording links, not live microphone dictation.

Can I upload MP3, M4A, WAV, FLAC or OGG?

Yes, these common recorded audio formats are supported. The file still has to satisfy size, per-job and account limits. If the media is on a public supported platform, the main app offers a URL workflow.

What are the LongScribe upload size limits?

Is LongScribe audio to text free?

The normal Free plan has 60 minutes across 20 recording jobs per month. Eligible short anonymous previews may be available; longer or repeated work can require sign-in or a paid plan. Free is not unlimited.

Can I transcribe a long audio recording?

Long recordings can process in the background once upload has finished and the job is accepted, but per-job limits, monthly transcription minutes and file size still apply. Check your plan first.

Does a two-hour MP3 recording fit the Free plan?

A two-hour MP3 uses 120 transcription minutes. It fits the monthly Free allowance only when at least 120 of 60 minutes and one of 20 recording jobs remain. The file must also fit upload-size and per-job rules. For recurring work, Premium is $4.99/month for 5,000 minutes and 500 monthly recording jobs.

Should I leave the browser tab open during transcription?

Keep the tab open while your file uploads. Once the job is accepted and processing, you can leave the page and return later to check the transcript.

Can I change transcript timestamps?

The text editor changes words, not the original start/end times of speech segments. SRT/JSON exports may preserve existing boundaries when a manual edit remains aligned to those segments.

How accurate is automatic audio transcription?

Accuracy varies with noise, language, names, accents and overlapping speakers. Review critical quotes, figures and terminology against the recording; AI text is not a human-certified transcript.

Can I download a Word file or SRT subtitles?

Yes. Completed transcripts can be exported as DOCX, TXT, JSON or a separate timed SRT subtitle file for compatible video editors.

Can I transcribe a private recording or public URL?

An authorized original audio file can be uploaded. Public source links have separate access checks and are not a way to bypass private or restricted media. Do not upload recordings without consent or appropriate rights.

Can AI transcribe noisy audio or a strong accent accurately?

Can it reliably transcribe English mixed with another language?

Multiple supported languages do not prove accurate switching within one recording. LongScribe has no published frozen code-switching benchmark; select the correct source language where possible and check language transitions against the audio.

Convert audio to text →

Convert long audio to text.Keep the timestamps.