Transcription: convert audio to text with AI
Chat shortcut: /transcribe
Upload a recording and have it transcribed: the AI turns speech into text that you can copy, save as a file, or summarize and translate via “Continue in Chat”. Transcription requires an account.
With Account
Capture what was said without taking notes
Typing up an hour of interview by hand takes half a working day. AI transcription does the job for you: you upload the recording, the model recognizes the speech and returns the wording as text, with punctuation and paragraphs, so it can be read right away.
You need that for interviews and research, for lectures and seminars, for meetings that exist as a recording, or for long voice messages you'd rather read than listen to. Your own dictations or podcast episodes become searchable and quotable this way too.
“Continue in Chat” brings the text into the chat, where you can have it summarized, turned into minutes with decisions and action items, or translated into another language, or ask questions about the content, all without uploading the file again.
How it works
- 1
Upload recording
Choose an MP3, WAV, M4A, OGG, WebM, MP4, AAC or FLAC file up to 25 MB.
- 2
Convert to text
Click “Convert to text” and wait until the transcript appears; the time depends on the length of the recording.
- 3
Use the text
Copy the transcript, save it as .txt or continue working with it in the chat.
Typical uses
Transcribe an interview
Upload the recording, proofread the transcript, and pull out the key quotes in the chat.
Meeting minutes
Transcribe the meeting recording and organize it into decisions and action items in the chat.
Review a lecture
Convert the recording into text and create a summary or flashcards from it.
Read a long voice message
Upload OGG or M4A and skim the content as text.
Make a podcast searchable
Transcribe an episode and turn it into quotes, show notes or a blog post.
Good to know
- Input: MP3, WAV, M4A, OGG, WebM, MP4, AAC and FLAC up to 25 MB; as an MP3 at 128 kbit/s that is roughly 25 minutes; other bitrates allow more or less.
- Output: flowing text to copy, as a .txt file or for further work in the chat.
- Billed in tokens by the length of the recording and the active transcription model; an account is required.
- The recording is sent to the transcription provider (ElevenLabs or OpenAI) for conversion; what it stores is governed by its terms. ChatX receives the file only to process it.
- You can upload MP4 videos directly; convert other video formats into an audio file first, for example with “Convert MP4 to MP3”.
- Best results with clear speech, little background noise and one speaker at a time.
Frequently asked questions
Which formats are supported?
MP3, WAV, M4A, OGG, WebM, MP4, AAC and FLAC up to 25 MB. Voice messages from WhatsApp or an iPhone usually come as OGG or M4A and work directly.
How long can the recording be?
The limit is 25 MB. How many minutes that is depends on format and bitrate: as MP3 at 128 kbit/s about 25 minutes, as WAV only a few minutes. Convert long WAV recordings to MP3 first with “MP3 Converter”.
Can I transcribe a video too?
You can upload MP4 files up to 25 MB directly. Convert other video formats or larger videos into an audio file first with “Convert MP4 to MP3”.
How accurate is the transcript?
With clear speech and little background noise, usually well readable. Errors are most likely with proper names, technical terms, strong accents and when several people talk at once. You should proofread important passages.
What can I do with the text afterwards?
Copy it, save it as .txt or use “Continue in Chat” to have it summarized, turned into minutes or translated, or to pull out action items. Working in the chat is billed by usage like a chat message; for very long transcripts the character limits per message apply.
What does transcribing cost?
Usage is billed in tokens and depends on the length of the recording and the active transcription model. You need an account for it.
What happens to my recording?
It is sent to the transcription provider (ElevenLabs or OpenAI) for conversion; what it stores is governed by its terms. ChatX receives the file only to process it; the transcript appears on the page.