--- title: "Audio to SRT: Free Subtitles from Audio | audiogram.video" description: "Turn an MP3, WAV or voice recording into an SRT or VTT subtitle file: speech written out with timings, editable line by line, then downloaded. Free, no sign-up." url: https://audiogram.video/audio-to-srt --- Free · no watermark · in your browser # Audio to SRT: make a subtitle file from a recording Choose up to 10 minutes of a recording, let speech recognition write it out with timings, correct it, and download an SRT or VTT file. To turn audio into an SRT file, open the recording above, choose the part you need, press **Make captions** and then download **SRT** from the **Captions** panel. Each line of speech becomes a numbered cue with a start and end time, which YouTube, video editors and media players read as subtitles. The words are written out by a speech-recognition model, so the clip’s sound has to be sent to our speech provider when you press the button. Only the part you selected goes, compressed, and neither we nor the provider keep it. Everything else, including your edits and the rest of the recording, stays in your browser. Before downloading, you can read every line and type over mistakes. You can also turn the same captions into a [captioned podcast clip](https://audiogram.video/podcast-clips). ## How to convert audio to SRT 1. ### Open the recording Press **Choose a recording** (Ctrl+O or ⌘+O). MP3, WAV, M4A, FLAC, OGG and voice memos open directly, and so do video files, for their sound. 2. ### Choose the part to caption Captions are made for the clip shown on the timeline. Drag **Clip start** and **Clip end** to cover what you need, up to 10 minutes. A recording of 75 seconds or less is already selected in full. 3. ### Make the captions The **Captions** panel is open. Leave **Language** on **Detect the language** or pick one of 24 languages, then press **Make captions**. The clip’s sound is sent to be written out, and the timed cues appear in the panel when it is done. 4. ### Correct the text Click any cue and type over it. The time on its left plays from that point; the cross deletes a cue you don’t want. Press **Make captions again** if you move the clip. 5. ### Download SRT or VTT Press **SRT** or **VTT** above the list of cues. The times start at 0:00 at the beginning of your clip. ## SRT or VTT: which file you need Both are plain-text files of timed captions, and both can be opened in any text editor. The differences are small but they decide where a file is accepted. * **SRT (SubRip)** is the most widely accepted. Each cue is a number, a time range such as `00:00:04,120 --> 00:00:06,900` (a comma before the milliseconds) and one or two lines of text. YouTube, Facebook, LinkedIn, Premiere Pro, DaVinci Resolve and VLC all read it. * **VTT (WebVTT)** starts with the word `WEBVTT` and uses a full stop in the times: `00:00:04.120 --> 00:00:06.900`. It is the format web browsers use for subtitles on a website’s own video player. If you are unsure, download SRT. If you are adding subtitles to a video on your own web page, download VTT. ## How accurate is it, and how to make it better The captions come from Whisper large-v3 turbo, a modern speech-recognition model. On clear speech from one or two people it gets most words right, with punctuation and capital letters. Results drop with background music, crosstalk, heavy reverb and unusual names. ### Before you press Make captions * Pick the **Language** yourself when a recording mixes languages or starts with music; detection listens to the opening and can guess wrong. * Keep the clip to the speech you need. Silence and music at either end can produce stray words. * Sung vocals over a full band are hard for any speech model. Expect to correct a lot, or paste the lyrics and press **Use this text** instead. ### After * Read the list once with the sound on: press **Space** to play and follow the captions in the preview. * Fix names and terms by typing over the cue; the timing stays. * Delete cues for coughs, music or sounds you don’t want subtitled. ## Already have a transcript? If someone has typed out the recording, you don’t need speech recognition at all. Paste the text into the box in the **Captions** panel and press **Use this text**: the words are spread across the clip by length. That timing is only approximate, so it works best for short clips with steady speech. Nothing is sent anywhere when you do this. To fix an SRT or VTT file made elsewhere, press **Open SRT or VTT**, correct the cues and download it again. A file timed to the whole recording keeps its times; a file that starts at zero is lined up with your clip. ## Recordings longer than 10 minutes Captions cover the selected clip, and a clip can be up to 10 minutes. For a short interview or a single segment that is plenty. For an hour-long episode it is not the right tool, and we would rather say so: * If the recording is going on YouTube, upload it there and use YouTube’s own automatic captions, then correct them in YouTube Studio. * Video editors such as Premiere Pro and DaVinci Resolve Studio (the paid version) can transcribe a whole timeline and export SRT. * You can caption a long file here in parts, but each SRT starts at 0:00, so the parts need joining and re-timing afterwards. ## Audio to SRT: frequently asked questions ### How do I convert an MP3 to SRT? Open the MP3 here, drag the timeline handles around the part you need, press **Make captions** in the **Captions** panel, correct the text and press **SRT**. The file downloads as captions.srt. ### Is my audio uploaded? Only the part you choose, and only when you press **Make captions**. That clip’s sound is compressed and sent through this site to our speech-recognition provider, which writes it out and doesn’t store it. We don’t keep it either. The rest of the recording never leaves your device. ### Is this SRT generator free? Yes, with no account and no watermark. There are fair-use limits on how much speech can be captioned in a short time, and each clip can be up to 10 minutes. ### Which languages can it write out? Leave **Detect the language** selected, or choose from 24, including English, Spanish, French, German, Portuguese, Hindi, Arabic, Japanese, Korean and Chinese. It writes captions in the language that is spoken; it doesn’t translate. ### Can I get word-by-word timings? The SRT and VTT files have one time range per line, which is what players expect. Inside the studio, words light up one at a time with the **Highlight** caption style, using the model’s word timings where it gives them. ### How do I add an SRT file to a video? On YouTube, upload it in YouTube Studio’s subtitles section for that video. In an editor, import it as a caption track. To burn the captions into the picture instead, export the video from this studio with a caption style switched on. ### Can I make an SRT from a video file? Yes. Open the MP4, MOV, WebM or MKV here and its sound is used; the captions are made from that. ## More ways to turn sound into video * [ Free audiogram maker Turn a recording into a video with a moving wave, cover art and captions. ](https://audiogram.video/) * [ Vertical podcast clips with captions Captions-first 9:16 clips from an audio episode, for Reels, TikTok and Shorts. ](https://audiogram.video/podcast-clips) * [ Podcast audiogram maker Square episode promos with cover art, captions and a saved style for your show. ](https://audiogram.video/podcast-audiogram) * [ How to make an audiogram Choosing the moment, the design, the captions and the sizes, step by step. ](https://audiogram.video/how-to-make-an-audiogram) * [ How to start a podcast on YouTube Set up a YouTube podcast for an audio show, with RSS, full episodes and Shorts. ](https://audiogram.video/how-to-start-a-podcast-on-youtube) * [ How to add your own sound on TikTok Record it, add a voiceover, or upload it inside a video as an original sound. ](https://audiogram.video/how-to-add-your-own-sound-on-tiktok)