Upload your file
Choose a supported audio or video recording up to 100 MB.
Create an SRT subtitle file from an MP3 narration, podcast excerpt or lesson. Upload the MP3 track, review the recognized speech and download timed captions for the media you are preparing.
Long recordings can take a few minutes. Keep this tab open.
An MP3 track can be the source for captions when you already have the audio used in a video. AudText accepts the MP3 directly and uses returned speech timestamps for the subtitle file. You can also download a plain transcript for scripts or notes. Uploads are limited to 100 MB, with no account required.
Choose the same audio version that appears in your final video. A podcast excerpt with a removed introduction or an MP3 trimmed after transcription has a different timeline. If the track starts later in the video, adjust the downloaded subtitles in your editor before publishing.
Use the best original MP3 available. Repeated MP3 encoding and very low-quality compression can make consonants harder to recognize. Converting the MP3 to WAV cannot restore detail already lost. If speech is masked by music or several people talk at once, check those passages with playback.
The SRT groups recognized speech into timed caption cues. Automatic breaks are a starting point: check caption length, reading time and synchronization in the editor where you will use the SRT. A long pause or overlapping speech may need manual corrections.
The plain text editor helps prepare scripts and document exports, but its changes do not carry into SRT or VTT. Correct misheard words and adjust cue timing in a subtitle editor after downloading. Keep the MP3 alongside the SRT so that synchronization and wording can be checked together.
Choose a supported audio or video recording up to 100 MB.
The system detects language, speakers, words and sound events, then aligns them with timestamps.
Play it back, edit the text and download the format you need.
Read along with the recording, copy clean text, or export files for documents, captions and your own workflow.
Click any word to jump to its exact moment and follow the active phrase during playback.
Different speakers are identified automatically, while events such as laughter or applause can be marked when detected.
Automatic mode can recognize speech across 90+ languages; seven common languages also have manual presets.
00:12 — “Click any word to return to the matching moment in the recording.”
Raw uploaded media is deleted after processing. Transcript text and technical request data may be retained in protected operational logs for service reliability, abuse prevention and support.
Privacy and retention →Yes. Upload the MP3 audio recording and download SRT when timestamp data is available. The captions follow the MP3 timeline; check their alignment after importing them into a video.
No. It provides a separate downloadable SRT file. Import that file into your media player, subtitle tool or video editor according to that tool’s caption workflow.
Converting an existing MP3 to WAV does not restore lost speech detail. Upload the best original source you have. If a clearer original WAV is available, the audio-to-SRT tool can accept it directly.
Edits to plain text do not update the generated captions. Use word playback to check the speech, then correct the downloaded SRT in a subtitle editor.
Use the audio-to-SRT workflow with your original supported file. There is no need to convert a valid M4A or WAV recording to MP3 just to generate captions.
Yes. You can transcribe and download results without a subscription or account.
MP3, WAV, M4A, MP4, FLAC, AAC, OGG, OGA, OPUS, WebM, WMA, AVI, MKV, MOV, MPEG and MPGA are accepted up to 100 MB.
Automatic detection covers more than 90 languages. English, Spanish, Portuguese, German, French, Russian and Turkish can also be selected manually.
Speaker information is added automatically. Sound events such as laughter or applause may also appear when detected; structured details are included in JSON.
Yes. Every result includes both SRT and VTT downloads with timestamps.
Accuracy depends on recording quality, background noise, compression, accents and overlapping speech. Verify important names, numbers and quotations against the recording.