XSpeak becomes better. Fully on-device, made to help you in real-time during conversations. Now with playback and more.
Hey folks, It's been around 4 months since my last post, and I worked on XSpeak during all this time. I'd like to share updates with you. But first, let me briefly describe the app. Problem Communication is sometimes challenging and have an effect on our lives. Just a couple of examples: Communication anxiety that makes it hard to start talking Employee talking to their boss and agreeing on something under pressure, realizing this only afterward Speaking with people and forgetting something you know you remembered XSpeak is made to help in these and similar situations. Compare Compared to alternatives, for example Fireflies, XSpeak uses on-device AI models to provide help. This means your conversations stay with you, which is very nice for a regular user, and the only possible way to record in some areas. Pricing $3.99 / Month (7 days free trial) $19.99 / Year (7 days free trial) $49.99 Lifetime New features Audio playback: you can now replay audio for any transcribed fragment Transcript editing: if recognition failed (happens), you can just relisten and fix it Ask anything in transcript: you can write your custom question or request Multi-speaker identification: once you name speaker it'll be remembered and recognized in future conversations Organize transcripts in folders, and more I know XSpeak already helps some of you. I hope it'll make more people happier in communication. 👉 Get XSpeak on App Store Happy to hear your feedback anytime, it helps. Thanks!
Collected discussion
Sounds great, I will look into that!
Thank you! Will be happy to hear any of your feedback :)
I’ve noticed a significant decline in transcription quality over time during meetings. After about 20-30 minutes, the quality becomes quite poor. Could it be that the context size grows too much, leading to this drop in quality? Also the drop offs increase (i.e., XSpeak doesn’t transcribe portions of the conversation). I am using an M3 Pro 36 Gb RAM machine, on MacOS 26
Hey, thanks for letting me know, I'm not aware of such issue. Couple of details that would help me to identify the issue: - What kind of mic do you use? - Do you record from system audio or not? - What language you record? English? - Do you hear any difference in audio quality if you play fragments in the beginning of the recording and in the end? I'll prioritize this and do my best to resolve in the next release. It's unlikely because of the context, since the context for transcription is reset every 5 minutes to avoid such issues. So it's something else. Edit: the problem is identified and the fix is submitted to App Store review. It's a problem with system audio mixing in long meetings. Edit 2: the fixed version is live on App Store
I use the system audio and use the app to transcribe Zoom meetings - all in English . The audio quality is good in general, and my mic is a wireless headset and sounds just fine. The last time I used the app was on a 1-hour meeting and towards the end of the meeting the transcription quality was terrible (some sentences read like “diarrhea octopus agree”). To the point that I found it unusable to generate the meeting minutes
This can be really usefull. I was using tl;dv but it's expensive and not verty good doing the job.
Thanks! Would be useful to hear about your experience. I’ll be happy to improve the app if something is not good enough for you.
This comment was deleted at the source.
Hey, thanks, I'm happy to hear you like it! I'm doing my best to make it beautiful and easy to use. Full audio playback along with export is high priority feature. It's not possible via the app now (you can only find audio files manually in ~/Library/Containers/com.tseiler.XSpeak/Data/Documents/chats), but I'm in progress of implementation. I hope it'll be released in one of the next ~3 updates.