Swords, a Windows desktop app GUI for Whisper #2605
Replies: 12 comments 12 replies
-
|
Windows Defender blocks execution and flags as containing a Trojan - Trojan:Win32/Sabsik.FL.A!ml |
Beta Was this translation helpful? Give feedback.
-
|
same her |
Beta Was this translation helpful? Give feedback.
-
|
Some Antivirus Vendors may flag the compile binaries as malware. We are working hard to solve it. |
Beta Was this translation helpful? Give feedback.
-
|
We have submitted this executable file (i.e. swords. exe) to Microsoft Security Intelligence for malware analysis. Submission details: Analyst comments: At this time, the submitted files do not meet our criteria for malware or potentially unwanted applications. The detection has been removed. Please follow the steps below to clear cached detections and obtain the latest malware definitions.
Alternatively, the latest definition is available for download here: https://docs.microsoft.com/microsoft-365/security/defender-endpoint/manage-updates-baselines-microsoft-defender-antivirus Thank you for contacting Microsoft. |
Beta Was this translation helpful? Give feedback.
-
|
No CUDA GPU's are available! gives a warning |
Beta Was this translation helpful? Give feedback.
-
|
It is recommended to utilize CUDA GPU's for faster processing instead of relying solely on CPU. |
Beta Was this translation helpful? Give feedback.
-
|
July 5, 2025: Swords v1.0.2 has been released. Version Update Notes You can download it here: Download |
Beta Was this translation helpful? Give feedback.
-
|
Hi. Tks for app. What model can recognize native languge as native and english as english in one record? |
Beta Was this translation helpful? Give feedback.
-
|
Whisper can handle paragraph-level mixed-language audio (such as multiple people using different languages in a meeting), but due to the 30-second slicing mechanism, its support for short sentence alternation is relatively weak. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks. This workaround is good for conversation of different languages persons. But i mean for example sentences with words on different languages. Like IT terminology. |
Beta Was this translation helpful? Give feedback.
-
|
Setting a list of professional terms in the initial_prompt parameter and setting the carry_initial_prompt parameter to "true" can guide the model to generate transcription results that are more suitable for academic lecture scenarios. |
Beta Was this translation helpful? Give feedback.
-
|
July 15, 2025: Swords v1.0.3 has been released. Version Update Notes You can download it here: Download |
Beta Was this translation helpful? Give feedback.
-
|
Hi there, I've installed the Swords 1.0.3 version, but when I transcribe the audio file I see that CPU is used, whereas CUDA of my GPU is supposed to be used instead, how can I fix this? |
Beta Was this translation helpful? Give feedback.
-
|
Swords 1.0.3 comes with OpenAI Whisper 20250625 built-in, which requires support from PyTorch 2.7.0 and CUDA 11.8. |
Beta Was this translation helpful? Give feedback.
-
So, I have both pytorch and CUDA with appropriate versions. I'm sharing the screenshot of my parameters, so u can see) |
Beta Was this translation helpful? Give feedback.
-
|
Troubleshooting:
Furthermore, we have observed that in certain circumstances, Swords 1.0.3 does not fully utilize the capabilities of the CUDA GPU. The exact reasons for this are currently under investigation. The next version will address this issue. |
Beta Was this translation helpful? Give feedback.
-
|
Swords 1.1.0 fully utilizes the capabilities of CUDA GPUs. You can download it here: Download. |
Beta Was this translation helpful? Give feedback.
-
|
August 2, 2025: Swords v1.1.0 has been released. Version Update Notes You can download it here: Download |
Beta Was this translation helpful? Give feedback.
-
|
Why are you releasing this through a random mediafire link? Why is there no website, no repo, nothing. |
Beta Was this translation helpful? Give feedback.
-
|
It will take some time to complete the construction of the website or repository. |
Beta Was this translation helpful? Give feedback.
-
|
August 18, 2025: Swords v1.1.1 has been released. Version Update Notes You can download it here: Download-1 or Download-2 |
Beta Was this translation helpful? Give feedback.
-
|
August 28, 2025: Swords v1.2.0 has been released. Version Update Notes You can download it here: Download-1 or Download-2 |
Beta Was this translation helpful? Give feedback.
-
|
After downloading and decompressing Swords, the subfolder 'model' contain 'small' model data. If Swords needs to use other models during runtime, they will be automatically downloaded from the website 'openaipublic.azureedge.net'. You can also pre download and place they in the ’model' subfolder: Download. |
Beta Was this translation helpful? Give feedback.
-
|
October 18, 2025: Swords v1.3.0 has been released. Version Update Notes You can download it here: Download |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Swords, a Windows desktop app GUI for Whisper, transcribe and translate audio offline on your personal computer.
Swords is a powerful and lightweight Windows GUI application that brings OpenAI Whisper to your desktop. Whether you're a student, professional, content creator, podcaster, researcher, or simply need high-quality speech-to-text transcription, Swords makes it effortless to transcribe audio and video content quickly and accurately.
Unlike cloud service providers and other softwares, Swords is completely free and provides you with unlimited hours of transcription time. Don't worry about running out of credits or paying extra for overages.
Changelog
October 18, 2025: Swords v1.3.0 has been released.
Version Update Notes
1.3.0
1.Add batch-processing feature: Multiple audio or video files can be selected as input. After the system processes a file, it automatically starts processing the next file automatically.
2.Add auto-saving feature: After the system transcribes a file, it automatically saves the output content to the folder where the input file is located.
3.Upgrade the built-in PyTorch in the CPU-Only version to 2.9.0.
4.Fix the discovered errors.
August 28, 2025: Swords v1.2.0 has been released.
Version Update Notes
1.2.0
1.Add the feature of setting "Advanced options"(i.e. beam_size, best_of, compression_ratio_threshold, logprob_threshold, no_speech_threshold, condition_on_previous_text).
2.Upgrade the built-in PyTorch to 2.8.0+cu126.
3.Fix discovered errors.
August 18, 2025: Swords v1.1.1 has been released.
Version Update Notes
1.1.1
1.Upgrade the built-in PyTorch in the CPU-Only version to 2.8.0.
2.Fix discovered errors.
August 2, 2025: Swords v1.1.0 has been released.
Version Update Notes
1.1.0
1.Fix the error: In some cases, Swords did not fully utilize the capabilities of CUDA GPUs.
2.Release the CPU-Only version, suitable for scenarios where CUDA GPU is unavailable.
3.Fix discovered errors.
July 15, 2025: Swords v1.0.3 has been released.
Version Update Notes
1.0.3
1.Fix the error: When CUDA GPUs are available, Swords immediately exits.
2.Add the feature of setting the 'Carry initial prompt' parameter.
3.Fix discovered errors.
July 5, 2025: Swords v1.0.2 has been released.
Version Update Notes
1.0.2
1.Fix the error: When you click the "Save as SRT File" button, the result you get is still a Text file.
2.Built-in OpenAI Whisper version: 20250625.
June 2, 2025: Swords v1.0.1 has been released.
You can download it here: Download
We hope you enjoy it and let us know if there are any other features you would like to see. You can email us at [email protected]
Beta Was this translation helpful? Give feedback.
All reactions