# Transcribe speech in audio or video files

Robot: `/speech/transcribe`

🤖/speech/transcribe transcribes speech in audio or video files.

You can use the text that we return in your application, or you can pass the text down to other <dfn>Robots</dfn> to filter audio or video files that contain (or do not contain) certain content, or burn the text into images or video for example.

Another common use case is automatically subtitling videos, or making audio searchable.

Set `speaker_labels` to `true` when you want JSON or meta transcription output to distinguish
recurring speakers:

```json
{
  "steps": {
    "transcribed": {
      "use": ":original",
      "robot": "/speech/transcribe",
      "provider": "aws",
      "format": "json",
      "speaker_labels": true,
      "max_speakers": 3
    }
  }
}
```

Speaker labels are currently supported by the `aws` and `gcp` providers. If you enable
`speaker_labels` without setting `provider`, Transloadit uses `aws` for that <dfn>Step</dfn>. Labels
are normalized as `speaker_1`, `speaker_2`, and so on:

```json
{
  "text": "Hello there. Hi!",
  "words": [
    { "text": "Hello", "startTime": 0, "endTime": 0.5, "speaker": "speaker_1" },
    { "text": "there", "startTime": 0.6, "endTime": 1, "speaker": "speaker_1" },
    { "text": "Hi!", "startTime": 1.2, "endTime": 1.8, "speaker": "speaker_2" }
  ],
  "segments": [
    { "text": "Hello there", "startTime": 0, "endTime": 1, "speaker": "speaker_1" },
    { "text": "Hi!", "startTime": 1.2, "endTime": 1.8, "speaker": "speaker_2" }
  ]
}
```

Stage: ga

## Usage example

Transcribe French speech from uploaded audio or video, and save it to a text file:

```json
{
  "steps": {
    "transcribed": {
      "robot": "/speech/transcribe",
      "use": ":original",
      "provider": "replicate",
      "target_language": "french",
      "format": "text"
    }
  }
}
```

## Parameters

* `interpolate`: Controls whether Assembly Variables are interpolated for individual instruction fields.

  By default, most Robot instruction fields interpolate Assembly Variables. Set this to `false` to treat every instruction field as literal text, or set an individual field path to `false` to treat only that field as literal text. For Robot-specific fields that are literal by default, set this to `true` or set that field path to `true` to opt back into interpolation.

  Use field names such as `path`, or dotted paths such as `ffmpeg.vf` for nested objects.

* `output_meta`: Allows you to specify a set of metadata that is more expensive on CPU power to calculate, and thus is disabled by default to keep your Assemblies processing fast.

  For images, you can add `"has_transparency": true` in this object to extract if the image contains transparent parts and `"dominant_colors": true` to extract an array of hexadecimal color codes from the image.

  For images, you can also add `"blurhash": true` to extract a [BlurHash](https://blurha.sh) string — a compact representation of a placeholder for the image, useful for showing a blurred preview while the full image loads.

  For videos, you can add the `"colorspace": true` parameter to extract the colorspace of the output video.

  For videos, you can also add `"interlaced": true` to detect whether the video is interlaced. This combines the cheap ffprobe `field_order` flag with a bounded `idet` sampling pass over the first frames of the source, exposing `interlaced`, `field_order`, and a diagnostic `interlace_detection` object under `file.meta`. This is computationally expensive and billed accordingly.

  For audio, you can add `"mean_volume": true` to get a single value representing the mean average volume of the audio file.

  You can also set this to `false` to skip metadata extraction and speed up transcoding.

* `user_meta`: Adds custom metadata to each file emitted by this Robot without modifying the file’s contents.

  The values are merged with any existing `user_meta` carried by the input file. If both objects contain the same key, this Robot’s value takes precedence. Assembly Variables are supported, for example `{ "internal_file_id": "${file.id}" }`.

* `result`: Whether the results of this Step should be present in the Assembly Status JSON

* `queue`: Setting the queue to 'batch', manually downgrades the priority of jobs for this step to avoid consuming Priority job slots for jobs that don't need zero queue waiting times

* `force_accept`: Force a Robot to accept a file type it would have ignored.

  By default, Robots ignore files they are not familiar with.
  [🤖/video/encode](/docs/robots/video-encode.md), for
  example, will happily ignore input images.

  With the `force_accept` parameter set to `true`, you can force Robots to accept all files thrown at them.
  This will typically lead to errors and should only be used for debugging or combatting edge cases.

* `ignore_errors`: Ignore errors during specific phases of processing.

  Setting this to `["meta"]` will cause the Robot to ignore errors during metadata extraction.

  Setting this to `["execute"]` will cause the Robot to ignore errors during the main execution phase.

  Setting this to `true` is equivalent to `["meta", "execute"]` and will ignore errors in both phases.

* `use`: Specifies which Step(s) to use as input.

  * You can pick any names for Steps except `":original"` (reserved for user uploads handled by Transloadit)
  * You can provide several Steps as input with arrays:
    ```json
    {
      "use": [
        ":original",
        "encoded",
        "resized"
      ]
    }
    ```
  * You can also tag input Steps with `as` to pass semantic intent to robots:
    ```json
    {
      "use": [
        {
          "name": ":original",
          "as": "image"
        },
        {
          "name": ":original",
          "as": "mask"
        }
      ]
    }
    ```

  > [!Tip]
  > That's likely all you need to know about `use`, but you can view [Advanced use cases](/docs/topics/use-parameter.md).

* `provider`: Chooses the best provider based on your request.

  Set this to `"aws"`, `"gcp"`, or `"replicate"` to force a specific provider.
  When set to `"auto"`, Transloadit uses `"replicate"` by default and uses `"aws"` when
  `speaker_labels` is enabled.

* `granularity`: Both `"full"` and `"list"` currently return the same complete transcription response. `"list"` remains accepted for backward compatibility.

* `format`: Output format for the transcription.

  * `"text"` outputs a plain text file that you can store and process.
  * `"json"` outputs a JSON file containing timestamped words. When `speaker_labels` is enabled, words can include `speaker` labels and the JSON can also include grouped `segments` by speaker.
  * `"srt"` and `"webvtt"` output subtitle files of those respective file types, which can be stored separately or used in other encoding <dfn>Steps</dfn>.
  * `"meta"` does not return a file, but stores the data inside Transloadit's file object (under `${file.meta.transcription.text}`, `${file.meta.transcription.words}`, and, when speaker labels are available, `${file.meta.transcription.segments}`) that's passed around between encoding <dfn>Steps</dfn>, so that you can use the values to burn the data into videos, filter on them, etc.

* `speaker_labels`: When enabled, Transloadit asks the transcription provider to distinguish different speakers. JSON and meta output can then include `speaker` labels such as `"speaker_1"` on individual words, plus grouped `segments` by speaker. Text, SRT, and WebVTT output behavior is unchanged.

  Speaker labels identify recurring voices, not real person names. Accuracy depends on audio quality, background noise, overlapping speech, and the number of speakers.

* `max_speakers`: The maximum number of speakers to detect when `speaker_labels` is enabled.

* `source_language`: Sets the spoken language for the `"aws"` and `"gcp"` providers. The transcription stays in the spoken language; this parameter does not translate speech.

  The language should be specified in the [BCP-47](https://www.rfc-editor.org/rfc/bcp/bcp47.txt) format, such as `"en-GB"`, `"de-DE"` or `"fr-FR"`. Please also consult the list of supported languages for [the `gcp` provider](https://cloud.google.com/speech-to-text/docs/languages) and [the `aws` provider](https://docs.aws.amazon.com/transcribe/latest/dg/what-is-transcribe.html).

* `target_language`: Sets the spoken language hint for the `"replicate"` provider. Omit this parameter to detect the spoken language automatically. Set it to the full lowercase English language name, such as `"french"`, when the spoken language is known. Despite its name, this parameter does not translate speech.

  For the `"aws"` and `"gcp"` providers, use `source_language` instead.