Telegram Bot Converts Voice Messages Into Text

This is the simple bot that converts voice into text. I tried to review available public speech recognition services and the results you can see below in the table.

Tech Stack

• Typescript
• ExpressJS
• Axios
• PostgreSQL
• Google Cloud API
• Google Analytics

Service model

flowchart BT
  subgraph tg[Telegram]
    voice[Voice message]
    audio[Audio]
    video[Video note]
    text[Text message]
    bot[AudioMessBot API]
  end
  subgraph cluster[Replicas]
    r1{{Replica 1}}
    ar{{Active replica}}
    r2{{Replica N}}
  end
  voice-->bot
  audio-->bot
  video-->bot
  bot-->text
  bot---ar
  ar---db[(PSQL\nDatabase)]
  ar---cloud((Cloud API provider))

Services overview

Service provider	Russian lang	Synchronous API	Duration limitation	File upload	Speed
IBM Watson	no	no	N/A	Unknown	Unknown
Microsoft Azure	no	no	N/A	Unknown	Unknown
Amazon AWS	yes	no	Unlimited	S3	Minutes
Google Cloud	yes	yes	1 minute*[1]	Direct / GDrive	Instant*[2]
Wit.ai	yes	yes	5 minutes	Direct	Instant

• For direct upload
  1 Unlimited for asynchronous upload via Google Drive
  2 Takes a while for asynchronous upload via Google Drive