Sample Voice Bot

Telegram text-to-speech bot. It works in both inline and text mode and offers a convenient interface to generate voice from the text.

Try it out: @samplevoicebot.

Requirements

Bot requires an AWS account to work. It uses Amazon Polly to generate voice and S3 to store and share generated voice files.

Configuration

Bot is configured via environmental variables, listed in config.py.

The table below shows required parameters:

Parameter	Description
TELEGRAM_BOT_TOKEN	Token issued by @BotFather
AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY	AWS credentials
AWS_S3_BUCKET	AWS bucket name where files will be stored
AWS_REGION_NAME	AWS region name where your bucket is located e.g. us-east-1

Other parameters:

Parameter	Values	Default	Description
TELEGRAM_ADMIN_ID	int	-1	User ID - admin of this bot
TELEGRAM_MIN_MESSAGE_LENGTH	int	1	Minimal message length to generate voice
TELEGRAM_MAX_MESSAGE_LENGTH	int	255	Maximum message length to generate voice (inline messages cannot exceed 256 characters)
TELEGRAM_INLINE_DEBOUNCE_MILLIS	int	1000	Debounce interval for incoming inline requests to avoid unnecessary voice generations
LANGUAGE_DETECT_MAPPINGS	str		Language mappings in format from=to. Multiple substitutions can be joined with commas e.g. mk=ru,bg=ru,uk=ru
MAX_WORKERS	int	4	A number of workers used for concurrent jobs (e.g. speech synthesis)
PREFETCH_LANGUAGES	str		List languages (e.g. ru,en) to prefetch voices list* on bot startup
VOICES	str		Specify voices to use, e.g. ru-RU=Tatyana,Maxim. Join multiple mappings with ;. List of supported voices
DEBUG	1, 0	0	Enable/disable verbose logging

* In inline mode bot uses only 2 voices per language to minimize synthesize requests

AWS configuration

An unrestricted, unauthorized public access to the objects in the bucket is required. This can be achieved with bucket policy:

Bucket policy

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "PublicRead",
            "Effect": "Allow",
            "Principal": "*",
            "Action": "s3:GetObject",
            "Resource": "arn:aws:s3:::samplebucket/*"
        }
    ]
}

* samplebucket must be substituted with your backet name.

See docs for more info.

Docker

Images

Images are available at ghcr.io.

Build image

$ docker build -t samplevoicebot .

Run image

Create a file named .env and put your credentials, for example:

AWS_ACCESS_KEY_ID=<your-aws-access-key-id>
AWS_SECRET_ACCESS_KEY=<your-aws-secret-access-key>
AWS_REGION_NAME=<aws-region-name>
AWS_S3_BUCKET=<your-s3-bucket-name>
TELEGRAM_BOT_TOKEN=<your-telegram-bot-token>
<any-additional-parameters>

Run:

$ docker run -d --env-file .env samplevoicebot