Main points from audio: how to get different summary variants
The 'Main points from audio' tool recognizes speech in the browser and then uses AI to highlight the main points and further actions. Often users get one summary variant and stop there. But if you change the goal formulation, you can get several variants and choose the most suitable one. In this article we will examine how to change criteria, compare three variants and refine the chosen one. All examples are illustrative.
Open tool: Audio summary ↗How the tool works
Upload an audio file (up to 20 MB, duration up to 5 minutes). Speech is recognized locally in the browser, i.e. the audio is not sent to the server. Then the text is passed to AI, which forms a summary and a list of actions based on the goal you specified.
There is a 'Goal or desired result' field in the form. It determines what the AI will focus on. If the goal is formulated generally, the summary will be superficial. If specifically - AI will highlight the necessary details.
After generation you get one result in a separate panel. It can be copied or downloaded in TXT format. Other formats, batch processing and automatic saving are not supported.
Example: upload a meeting recording, specify the goal 'Identify tasks for the marketing department for the week'.
Why one variant is just the beginning
AI gives one result at a time. If the goal is not specified, it selects an averaged variant that may not match your tasks. For example, with the goal 'Make a summary' you will get a general retelling where secondary details will be included.
To get different variants, you need to sequentially change the goal and generate the summary anew. Each time AI will emphasize accents differently. This is not an automatic enumeration, but a manual process, but it gives flexibility.
Important: the tool does not save history. If you changed the goal and generated a new variant, the previous result must be saved manually (copy or download), otherwise it will be lost.
First variant: goal 'General summary'. Second: goal 'Only solutions and responsible persons'. Third: goal 'Questions requiring clarification'.
How to change criteria to get different variants
Criteria are set via the 'Goal or desired result' field. The more specific the goal, the more focused the summary will be. For example, instead of 'Make a summary' write 'Identify three main problems raised by the client and proposed solutions'.
You can change not only the content, but also the format. Ask for 'List of actions with responsible persons and deadlines' or 'Brief summary in three sentences'. AI adapts the output to the request.
Another technique is to change the level of detail. Goal 'Only key numbers and facts' will give one result, and 'Detailed retelling with examples' - another. Experiment with formulations to find the optimal variant.
Goal 1: 'Identify budget questions'. Goal 2: 'Task list for designer'. Goal 3: 'Risks and unresolved questions'.
Comparing three variants: practical example
Suppose you have a recording of a meeting with a client. You want to get three different summaries. First, specify the goal 'General meeting summary' - you will get a retelling of all topics. Then 'Only agreements and next steps' - you will get a list of actions. Then 'Client objections and our answers' - you will get an analysis of the discussion.
Compare three results. The first may be useful for a report, the second - for planning, the third - for training employees. Choose the one that better corresponds to the current task.
If none of the options suited, refine the goal. For example, add 'With emphasis on deadlines' or 'Without technical details'. Remember that each time a new result is generated, and it needs to be saved separately.
Comparison: Option 1 - 5 topics, Option 2 - 3 tasks with responsible persons, Option 3 - 2 objections and answers.
Selection and refinement of the best variant
Having chosen the best variant, check it for accuracy. AI may make mistakes: miss important things, misunderstand context or add extra things. Compare with the original recording, especially in key moments.
If necessary, edit the text manually. Copy the result to a text editor and make edits. The tool does not allow editing inside the panel, but you can download TXT and work with it.
For refinement you can generate another variant with an clarified goal. For example, if the selected summary lacks deadlines, specify the goal 'Task list with dates' and get a new result. Compare it with the previous one and combine the best manually.
After checking you add a missed deadline to the summary and remove an insignificant remark.
Limitations and important nuances
The tool does not support batch processing, automatic saving, export to DOCX or PDF, integrations and publishing. Everything is done manually: loading, generation, copying or downloading.
Audio is recognized locally, but the resulting text is passed to AI for summarization. Do not upload confidential recordings if you do not want their content to be processed by a third-party model.
AI may make errors, and queue and availability of the service may change. Do not rely on the tool for critical decisions without verification. Always compare the result with the original audio.
If the recording contains background noise, recognition may be inaccurate, and the summary will require additional checking.
The 'Main points from audio' tool allows getting different summary variants if you change the goal in the 'Goal or desired result' field. Compare three variants, choose the best one and refine it manually. Remember the limitations: one result at a time, no auto-save, possible AI errors. Use the tool as an assistant, but always check the result.
Open tool: Audio summary ↗Interface translations are machine-assisted and may need refinement.