Add timestamping to speech to text

[msluszniak]

Description

Introduces a breaking change?

• Yes
• No

Type of change

• Bug fix (change which fixes an issue)
• New feature (change which adds functionality)
• Documentation update (improves or adds clarity to existing documentation)
• Other (chores, tests, code style improvements etc.)

Tested on

• iOS simulator
• Android simulator
• iOS device
• Android device

Testing instructions

Run demo app in apps/speech and run transcription for both timestamping and regular mode.

Screenshots

Related issues

Checklist

• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have updated the documentation accordingly
• My changes generate no new warnings

Additional notes

[chmjkb]

How about instead of a type union, returning a single type? I checked OpenAI docs and for word-level timestamping they're doing something like this:

{
  "task": "transcribe",
  "language": "english",
  "duration": 8.470000267028809,
  "text": "The beach was a popular spot on a hot summer day. People were swimming in the ocean, building sandcastles, and playing beach volleyball.",
  "words": [
    {
      "word": "The",
      "start": 0.0,
      "end": 0.23999999463558197
    },
    ...
    {
      "word": "volleyball",
      "start": 7.400000095367432,
      "end": 7.900000095367432
    }
  ],
  "usage": {
    "type": "duration",
    "seconds": 9
  }
}

This is likely familiar for the user, if he ever used the OpenAI API, and the user doesnt have to merge words by themselves when using timestamps.

[msluszniak]

Ok, so we want to always return plain transcription and additionally list of Words if needed, right?

[msluszniak]

And the second question is, Does OpenAI always return timestamps and full transcription, or is this optional as we have it right now?

[chmjkb]

I think it's optional, so we only return if needed. It makes sense to me to match entirely the structure they're returning