Typing out a recording by hand takes far longer than the recording itself. Speech to Text is an AI speech to text tool that does the first pass for you, so your time goes into checking and editing rather than pausing and rewinding.
You upload one audio or video file, in MP3, MP4, MPEG, MPGA, M4A, WAV or WebM format, up to 25 MB. The file goes to a speech recognition model and comes back as plain text in the language spoken in the recording. There is no prompt to write and no language to pick; the tool listens and writes down what it hears.
The transcript is the raw material for other work: a summary with the Text Summarizer, an article or a report. Speech to Text belongs to the Images, audio and video family, next to Text to Speech, which does the opposite job.
What you get
- A plain text transcript of the whole file, ready to copy or edit
- Text in the language spoken in the recording, with no translation step
- Support for MP3, MP4, MPEG, MPGA, M4A, WAV and WebM, so most phone and video recordings work as they are
- Video files accepted directly, the tool uses the audio track
What you fill in
- Audio or video file required
-
Clear audio with one person speaking at a time gives the best results. Split longer recordings into files under 25 MB.
How to use it, step by step
-
Check the file format and size
The tool accepts MP3, MP4, MPEG, MPGA, M4A, WAV and WebM files up to 25 MB. A long meeting recorded in WAV can pass that limit quickly, so export it as MP3 or M4A, or split it into parts.
-
Upload the file
Open Speech to Text in your workspace, drop the file in the upload area or click to browse, and start the transcription. You do not need to describe the content or choose a language.
-
Read the transcript against the audio
Listen to the passages that matter most, such as names, numbers and decisions, and correct them in the text. Speech recognition is reliable on clear speech and weaker on overlapping voices and background noise.
-
Reuse the text
Copy the transcript into your notes, or paste it into another tool of the suite to summarize it or turn it into a blog post.
Example, input and output
Input
- Audio or video file
- site-visit-notes.m4a, a two-minute voice note recorded on a phone
Output
Illustrative example.
Tips for better results
- Record close to the speaker. A phone on the table in a quiet room gives a cleaner transcript than a laptop microphone across the room.
- Keep music and notification sounds off while you record. Background audio ends up in the transcript as noise or garbled words, and it is easier to avoid than to correct.
- Split long recordings at natural breaks, such as agenda items, so each transcript stays easy to check.
Who uses it, and for what
Consultants after a client call
Record the call with the client's consent, transcribe it and keep a written record of what was agreed, ready to turn into a recap email.
Journalists and researchers
Transcribe interviews so quotes can be checked word for word, then mark the passages to use in the piece.
Video creators
Get the spoken text of a video as a starting point for subtitles, a description or a blog post based on the same content.
Limits and good practice
- Files are limited to 25 MB each. Longer recordings need to be compressed or split before upload.
- The transcript does not label who is speaking. With several speakers, add the names by hand where it matters.
- Names, figures and specialist terms can be misheard. Anything you publish or send to a client needs a check against the audio.
- The tool transcribes, it does not translate. To get the text in another language, run the transcript through the Translate tool.
Frequently asked questions
Which audio and video formats can I upload?
MP3, MP4, MPEG, MPGA, M4A, WAV and WebM, up to 25 MB per file. Most phone voice memos, meeting recordings and short videos fit these formats. If a file is too large, export it at a lower bitrate or split it into shorter parts.
Do I have to choose the language of the recording?
No. The transcript comes out in the language spoken in the recording. The tool does not translate, so an Italian interview gives an Italian transcript. Use the Translate tool afterwards if you need the text in another language.
Can it tell different speakers apart?
No, the output is continuous text without speaker labels. For interviews and meetings, add the names by hand while you review the transcript, or record each speaker on a separate track when that is possible.
How accurate is the transcript?
Accuracy depends mostly on the audio. One clear voice close to the microphone gives a transcript that needs few corrections. Background noise, music and crosstalk lower the quality. Always check names, numbers and decisions before you rely on them.
Is transcription included in every plan?
Transcription uses credits from your plan, counted on the words in the transcript, and each plan has a daily limit on transcription requests. Check the pricing page for what each plan includes before you plan a large batch of recordings.
Try it in your workspace
Start free, no credit card needed. When you need more, one suite grows with you, from Lite to Agency.
Sign in or create a free account, the tool opens in your workspace.