Magix Video Deluxe 2027

[Subtitles]

I ordered today Magix Video Deluxe 2027 Premium that is on sale these days.
Apart from the usual video editing features, what made me order it is that it has some nice new features:
AI Speech to Text Offline
AI Text to Speech
Facial Recognition
The license is for two computers.

The Speech to Text works reasonably well for videos (movies and TV shows), but if there is music background it will not work so well. Tested this with video songs and operas and I didn't get any text.
Also videos with multiple audio tracks seems to make it have a lot of errors.
In short, if you need accurate subtitle file, use other applications or use Magix Hub credits (2 hours included in the Premium)

[fred_abila]

Magix speech-to-text falling apart on music or a second audio track is a known limitation of most bundled recognizers: they transcribe the mix, not the voice. Split or mute the music stem before you run recognition, or duplicate the timeline and solo the dialogue track, then run speech-to-text on that. For accuracy after that, Subtitle Edit with its Whisper or forced-alignment tool usually beats Video Deluxe on messy audio, and you can import the .srt back if Magix will take it. If the deliverable is a talking-head or course video rather than a multi-track edit, a hosted caption pass is simpler. Disclosure: I build Vidmoat, a hosted AI video editor. You upload the video and ask it to caption from the speech, then fix wording in chat before export. It will not out-accuracy a careful Subtitle Edit pass on a film mix with music under the dialogue, but it skips the manual line-drag for straightforward voice tracks. For Video Deluxe projects with music beds, solo the voice first, then Whisper in Subtitle Edit.

[Subtitles]

Thanks for your reply.
I have several years of experience with the original Whisper AI cmd version so I know my way around.
I know that for best results and accuracy it is best to transcribe the speech portion using audio applications first.
I just wanted to check the Magix Deluxe AI Speech to text and see if it is worth using it because it is a GUI application, Unfortunately after several tests I can say it is not reliable enough. Also it defaults to translating to English instead of the original language (Italian, French, German, etc.) and I can't find a solution so that it transcribes the original language instead of translating to English.

[fred_abila]

The published Video deluxe help for speech to text does not show a transcribe-versus-translate switch. The control it documents is the language choice before you start.

Right-click the object, Speech to text (on a video object that is under Audio features), then set the language to the spoken one, Italian or French or German, instead of Auto Detect, and click Start analysis. After it finishes, the dialog shows the detected language. That page is the 2026 Hub flow. If 2027's offline tool looks different and there is still no language list, I would not keep hunting for a hidden toggle.

If you set the source language and the text still comes out in English, the tool is translating, and the help page does not expose Whisper's task switch. The command-line path you already know is the reliable one for the original language: whisper with --task transcribe and --language it (or fr, or de). That writes the source language rather than English. Import the .srt if Video deluxe will take it. Auto Detect plus an English UI is the combination that most often triggers a translate.

[Subtitles]

Thanks again for your reply. It is good that you have experience with Magix Deluxe.

The Magix Deluxe 2027 AI Speech to Test has been a big disappointment so far. It doesn't matter what language do I select to transcribe, it seems to have a mind of it's own.
An English movie with English audio gets me subtitles in Italian for example and it doesn't transcribe all the movie but just few parts.
Magix Deluxe doesn't allow importing a good srt file, so it is better to watch outside Deluxe.
Hopefully they can improve this in future updates.
It is a shame that they got this far, but still have these problems.
I haven't seen any other application that offers free offline AI Speech to Text. Filmora for example needs credits to work.