German Transcription: Video & Audio to Text

How German speech transcribes

German speech transcribes very well, including Austrian and Swiss Standard German. Spoken Swiss German dialect (Schweizerdeutsch) is a different matter and is recognised poorly.

What a German transcript looks like

German transcripts capitalise nouns and write compound words as single words, as in standard spelling. Long sentences with the verb at the end are kept intact, so you may want to break them up when quoting. Swiss German dialect is written out in Standard German spelling, with more errors than standard speech.

Examples

Costs: 90 credits (9¢) and 28 credits (2.8¢), including a translated transcript if you select another language. Audio files go in the audio to text workspace.

Sie or sie: one capital letter, three meanings

In speech, Sie (formal you), sie (she) and sie (they) sound identical. Only the verb form and context separate them, and sometimes not even those: "Kommen Sie morgen?" and "Kommen sie morgen?" are the same sounds. The transcript has to guess the capital. In interviews and customer calls where the speaker addresses someone formally and also talks about other people, check every sentence-internal sie and Sie, along with Ihr and ihr. A wrong capital doesn't look like an error, but it changes who did what. In start-ups and many workplaces people use du with each other, so a company video can switch between du and Sie depending on who is addressed; that switch is meaningful and should survive editing.

ß, ss and Swiss readers

Swiss Standard German doesn't use ß at all: Strasse, gross, Fuss. If the transcript is for Swiss readers, see which spelling appears; replacing ß with ss is a safe global find-and-replace for Swiss use, while the reverse isn't, because the choice depends on the preceding vowel. Readers in Germany and Austria expect ß after long vowels and diphthongs. Swiss vocabulary is a second layer: Velo for bicycle, parkieren for parken, Trottoir for pavement, Natel for a mobile phone. These are correct Swiss Standard German, so if the transcript replaces them with German-German words, restore them for a Swiss audience. The dialect question is separate: when a Swiss speaker drops from Standard German into dialect for an aside, mark the timestamps so someone who speaks it can check those lines.

Verbs that arrive in the next cue

German moves finite verbs and separable prefixes to the end of the clause: "Wir rufen Sie morgen nach der Besprechung mit dem Team zurück." In subtitles a sentence like this often spans two cues, and the viewer only learns the meaning (zurückrufen, to call back) at the very end. When editing the SRT, break at a clause boundary where you can, or shorten the first cue so the verb arrives sooner. The guide to adding subtitles to a video covers getting the edited file onto your platform.

Finding words inside compounds

Compounds such as Kundenzufriedenheitsumfrage or Datenschutzbeauftragte are written as one word, so a whole-word search for Umfrage or Datenschutz finds nothing. Turn off whole-word matching and search for the stem, remembering that the linking -s- or -n- (Arbeitsplatz, Sonnenschein) can sit between the parts. Also watch for compounds written differently from your house style (E-Mail-Adresse, E-Mail Adresse, Emailadresse) and for English verbs with German inflection, such as gedownloadet, gecancelt and geupdatet, whose spelling varies even among careful writers. Searching AI transcripts has more techniques.

Numbers spoken units-first

German says the units before the tens: einundzwanzig is one-and-twenty, siebenundneunzig is seven-and-ninety. That order is a common source of swapped digits when people take notes by hand, and it is worth checking in a machine transcript too, especially in prices, phone numbers read in pairs (null eins siebzig, dreiundzwanzig) and years (neunzehnhundertneunundachtzig). Speakers also say zwo instead of zwei to avoid confusion with drei on the phone, and Austrians may say Jänner for the month in dates. Figures in financial calls and technical training deserve a dedicated listening pass: search the transcript for digits and play each timestamp.

Costed: a 75-minute panel in Vienna

Source

75-minute MOV from the venue camera, Austrian and German panellists

Transcript run

75 credits (7.5¢), German selected (no translation)

English version

Second run with English as the target, another 75 credits (7.5¢)

Check first

Sie and sie, panellists' names and titles, Austrian words such as Jänner and Matura

Delivered

Timestamped transcript for the press office, German and English SRT for the recording page

The MOV to text page covers camera files like this one.

Using the tool on this page

• The spoken language is detected automatically; you select the output language — the spoken language itself for an untranslated transcript, or another language for a translation. Files can be up to 2 hours long.
• Only the audio is sent for processing — the video stays on your device — and audio and text are deleted within 30 minutes of delivery (how files are protected).
• You can download a timestamped transcript, subtitles (SRT and VTT) and plain text. Every output, step and limit is explained on the video to text page.