Text Analysis: topic, key points and tone of a recording
Add a recording or a link and Text Analysis starts by itself: it transcribes the speech and returns topic, up to 5 key points, tone, speech features and a summary.
Updated November 2, 20262 min read
Text Analysis suits a quick look at a recording: what a voice message from a client is really about, what tone a speech has, what a meeting was for. The result is five short sections that fit on one screen.
What you can analyze
| Source | Limit |
|---|---|
| Audio file (MP3, M4A, WAV, OGG, OPUS and other common formats) | up to 100 MB |
| Video file: the sound is extracted on your device | the sound up to 100 MB |
| Record audio | no time limit, file up to 100 MB |
| Paste link (YouTube, TikTok, Instagram and others) | up to 2 GB |
| Text from another tool: Analyze text under a transcript in Text Transcription or Video to Text | no file needed |
Text Analysis has no field for typing or pasting your own text. To analyze a text you already have, paste it into the chat and ask for an analysis, see Text Analysis or AI Chat.
How to run an analysis
Open the Tools tab and tap Text Analysis.
Add a recording: Record audio, Choose or Paste link.
Wait. First Recognizing speech..., and the transcript appears on the Text tab. Then Analyzing text..., and the result appears on the Analysis tab.
Both steps start by themselves; there is no Analyze button. The language of the recording is detected automatically.
What the analysis shows
| Section | What it contains |
|---|---|
| Topic | the main subject in one sentence |
| Key Points | up to 5 most important facts or ideas |
| Tone | one or two words, for example neutral, positive, anxious |
| Speech Features | repetitions, hesitations and other notable things in how the person speaks; may be empty |
| Summary | one or two sentences with the conclusion |
The analysis is written in the language of the app, even when the recording is in another language. It can be copied or shared. Under it, What next? has Discuss analysis, which opens the chat with the analysis already in it. The result is saved to the tool's History.
How much it costs
You pay for two things: speech recognition, about 0.73 tokens per minute with the default engine, and the analysis itself, which costs a fraction of a token for a typical recording. When the text comes from another tool, only the analysis is charged. Prices are in token prices for each tool.
When another tool fits better
For a long lecture or meeting you will want more than five lines. The Essence in Text Transcription or Video to Text gives chapters with timecodes, questions and answers and action items.
To ask follow-up questions or get the result in your own format, use the chat.
The sections are fixed and cannot be changed.
Contact support
Describe the problem and we will reply by email. Add the file name or job ID if you have one.