IGTranscript

Instagram transcript translator: transcribe first, translate second

They are two steps, and this tool only does the first one. Ask it for words and you get the words that were spoken — a Reel in French comes back in French, never in English — so translating means taking that text somewhere else afterwards. Doing it in that order is the steadier route for a concrete reason: the transcript is a file you can read and check against the audio before anything is translated, and a wrong word you can see is a wrong word you can fix, while a wrong word inside a translation is invisible because the sentence still reads smoothly. The ordered route also survives the failure this model cannot avoid — one language is chosen for the whole file and it does not detect switches — because a mixed Reel whose second half came back wrong is caught at step one, before the translation has baked the mistake into a second language.

InVideo WW

The rest of this page is the evidence: the two token ids in the speech model’s own configuration that decide transcribe versus translate, and why one of them can never be selected here; the three exported files and which of them survives a paste into a translation tool; what a second pass costs in text bytes when the transcript is 3,840,000 bytes of audio for a minute, and why the script is the cheapest input a translator can be handed; and the measured case where 19 of 338 words came back wrong, which is the check that has to happen before step two.

What the model’s own configuration says about translating inside it

Whisper does carry a translate task. Its configuration lists two task tokens and gives them ids, and the difference between them is one integer: task_to_id is {"transcribe": 50359, "translate": 50358} in both checkpoints this page uses. So nothing about the model prevents a translation; the question is whether the code path this page takes ever asks for one.

The decoder's first three positions. Position zero holds the start token, position one is left open for the language, position two is filled by the configuration with the transcribe id. The translate id sits one integer below it and no path reaches it. Three slots are filled before the audio decides anything Position two is the task, and it is written by the config, not chosen per run. decoder position 0 start token fixed 1 language left null in the config 2 task: transcribe forced to one id translate never selected one id away one integer apart in the same table 50359 transcribe 50358 translate 50259 to 50357 the 99 languages to listen for The output language is not a setting here. It is whatever position two was written with.
Why the transcriber cannot be used as a translator. The decoder is primed with three positions before it sees any audio: a start token at 0, the voice language at 1 with its value left open in the configuration, and the task at 2 where forced_decoder_ids [[1, null], [2, 50359]] writes the transcribe id. The translate task is not missing from the model — task_to_id gives it 50358, exactly 1 below transcribe — it is simply the id nothing here asks for. The 99 languages from 50259 to 50357 sit below both, because a language token says what the voice is, not what the output should be.

Three files, and which one survives a paste into a translator

The export is a dead file, which is exactly what a second step needs: nothing about it phones home, and every translation tool accepts text. What differs is how much of the file is not words.

The same transcript exported three ways, each a bar divided into words and non-word characters: plain text is almost entirely words, the timed text has a fixed stamp in front of every row, and the subtitle file repeats that stamp twice per cue with an index line added. Arrows point from each bar to a translator, and only the first bar has nothing in it that a translator should not touch. What a translator is handed, file by file Only the first bar is all words. The other two carry timing a translator cannot use. one 161.2-second run, 27 rows plain .txt words end to end paste with stamps clock on every row strip first .srt / .vtt clock before every cue not for text one row, as received [00:00:12.400] the words that were spoken A stamp cannot be mistranslated. The digits and colons around it can be pulled into the sentence. The subtitle shape compounds it: two timecodes and an index line per cue, before any word.
Three exports of the same 338-word run, and only one of them is a clean input. The plain text is 1,875 characters with the stamps switched off. Switching them on makes the block 2,280 characters, and 405 of them — 27 rows × 15 characters, a bracket, a 12-character timecode and a closing bracket — are clock, which is 17.8% of what gets pasted. The .srt and .vtt go further and carry 648 characters of timing, because each of the 27 cues opens with an index line and two 12-character timecodes joined by an arrow, 24 structural characters from the cues alone before the indexes are counted. A cue holds a start, an end and text and no language field, so a translated subtitle no longer lines up with the audio unless the timecodes were left exactly alone.

What the second pass costs, and why the script is the cheap end of it

The speech side of the run is expensive in bytes and the text side is not, which is the whole argument for doing the cheap step twice rather than the expensive one. The transcript is what you carry across; the audio does not go with it.

Why the order matters more when the audio is mixed

One language is chosen for the whole file. The picker holds 99 ids and one of them is applied to every 30-second window; the model does not detect a switch, and nothing in the pipeline compares window to window. A Reel that changes language half-way is therefore the case where a translation built on top of a bad transcript is worst: the wrong half comes back as confident nonsense in the source language, and the translation renders that nonsense fluently in the target language, where you can no longer tell it from a correct sentence.

Two routes to a translated Reel. On the top route the audio is transcribed once, checked against the audio, and only then translated, so a wrong word is caught before it is copied into a second language. On the bottom route translation and transcription happen in one pass, and the same wrong word arrives already fluent. Where the error becomes invisible Same mistake, two routes: one shows it, one buries it in a fluent sentence. transcribe, check, then translate audio one pass transcript read against the audio translation from known-good text one pass, output already translated audio one pass a fluent sentence carrying a wrong word reads as well as a correct one, so nothing flags it The check that costs one reading is only available on the top route. On the bottom route there is nothing left to check against.
Why the check has to sit between the two steps. On the top route the transcript is a file you can read against the audio before anything is translated, and the 19 wrong words measured in a 338-word run — 5.62%, in 11 places, 4 of them names — are visible there as words that do not match what you hear. On the bottom route the same mistake has already been rendered into the target language, where the sentence is grammatical and reads exactly like a correct one. The 405 stamp characters in the timestamped export and the 648 characters of timing in the subtitle files are the same problem from the other side: they are things a translator has to be told to leave alone rather than text.

What to do with the result, and what the tool will not do

Three shapes of answer, and they differ in where the second step happens rather than in whether it exists.

Can I translate an Instagram Reel without transcribing it first?

Not with this tool, and the reason is in the model rather than in the page. Its configuration pins the task at position 2 of the decoder to the transcribe id, and the translate id one integer below it is never selected, so every run returns the spoken language. A tool that offers translation in one step is doing the same two operations behind one button — it still has to recognise the speech before it can produce a sentence in another language, and it still cannot tell you which source word it got wrong.

Does the transcript come back in the language I selected?

Yes, and that is a limitation as much as a feature. The picker’s 99 ids are languages to listen for — English at 50259, Spanish at 50262 — not languages to write in, so a French Reel transcribed with French selected returns French words. There is no setting that would return English instead, which is why the translation is a separate step and why the wrong picker choice damages the transcript before any translation exists.

Can I translate the subtitle file instead of the plain text?

You can, but keep the timing lines out of the translator’s hands. A 27-cue .srt carries 648 characters of timing — an index line and two 12-character stamps per cue, joined by an arrow — and a translator handed the whole file will try to make sense of those lines along with the words. Translating only the text line of each cue and leaving the stamps alone preserves both the sync and the fixed 12-character field.

What about a Reel that has two languages in it?

Transcribe it first and look for the switch, because the export will not mark it. One language is set for the whole file, applied to every 30-second window without comparison between windows, and a cue carries a start, an end and text with no language field. A 120-second Reel is 6 windows; cutting at the switch and running each half with its own language costs one extra pass over one half and is the only way to get two transcripts a translation can rely on.

Why does the order matter if the translation reads fine?

Because “reads fine” is the failure mode. Measured on a 338-word English run, 19 words came back wrong — 5.62% — and 4 of the 11 places were names, which is the class of error a reader cannot repair from context. In the source language you can hear the difference; once those 19 words are translated into a sentence that is grammatical in the target language, nothing on the page says which of them was never spoken.

Last checked: 9 November 2026. The two token ids, task_to_id at {"transcribe": 50359, "translate": 50358} and forced_decoder_ids [[1, null], [2, 50359]], together with lang_to_id holding 99 ids from 50259 to 50357 and no_timestamps_token_id at 50363, were read from generation_config.json fetched from huggingface.co for onnx-community/whisper-base (3,832 bytes) and onnx-community/whisper-tiny (3,772 bytes) on this machine today; both files agree field for field. That no translate path is taken comes from app.js, which passes task: 'transcribe' and passes nothing else that could override it. The byte figures are arithmetic on constants in app.js and on the Web Audio spec — RATE = 16000, one channel, a 32-bit float sample, so 64,000 bytes a second: 60 seconds is 3,840,000 bytes and 120 seconds is 7,680,000. The 1,875 and 2,280 characters, the 405 stamp characters, the 648 timing characters, the 338 words, the 27 cues and the 19 wrong words in 11 places (4 names, 5.62%) come from the 161.2-second run recorded on this machine on 27 October 2026. The 404-word and 54-word counts are the same measurement record’s picker test. The 1194 words in 600.3 seconds is a run on this machine on 27 October; the resulting 59.67 words per 30-second window and the 358-character figure that follows are arithmetic on it and assume 6 characters a word, which is an assumption about English prose rather than a measurement. 6 windows for 120 seconds is arithmetic on the pipeline’s window of 480,000 samples and its jump of 320,000. The 79.7 MB and 43.6 MB model sizes are the decimal sizes this site states elsewhere, and 3,840,000 ÷ 40,844,231 = 9.40%, with 1,875 ÷ 43,600,000 = 0.0043%, are computed from the tiny checkpoint’s measured byte count. The two network requests are the dynamic import of the library and the fetch of generation_config.json, both in app.js.

If the trouble is the language you picked rather than the translation, picking Hindi and other non-English languages covers how one token is placed at the front of every window. If the Reel switches language part-way, the mixed-language page works out where the switch falls inside six windows. If you want the translated lines back in a file, the SRT and VTT page covers the 12-character stamp, and the repurposing page covers what else the same transcript can feed.