Wagtail-speech

Turn Wagtail pages into lifelike speech using Amazon Polly.

• Issues: https://github.com/moorinteractive/wagtail-speech/issues
• Testing: https://travis-ci.org/moorinteractive/wagtail-speech
• Coverage: https://coveralls.io/github/moorinteractive/wagtail-speech

Installation

Install the package

pip install wagtail-speech

Add wagtailspeech to your INSTALLED_APPS

INSTALLED_APPS = [
    # ...
    'wagtailspeech',
]

Now make and run your migrations

manage.py makemigrations
manage.py migrate

We assume you already have setup credentials for boto3.

Usage

Use the SynthesizeSpeechMixin on pages you want to be rendered as audio streams. These pages triggers the synthesize_speech_from_page method when they are edited via the Wagtail admin interface. You can find your saved audio stream on the page.audio_stream property (by default a low bitrate .mp3 file).

from wagtail.wagtailcore.models import Page
from wagtailspeech.models import SynthesizeSpeechMixin

class HomePage(SynthesizeSpeechMixin, Page):
    def get_speech_text(self):
        return self.title

You are completely free how to provide the text to be renderend. In most cases you probably also want to provide values from a StreamField. For this use case we provided a get_speech_text_from_stream_field method which calls get_speech_text (if present) on your blocks and concats the content of it with breaks between them.

class ExampleBlock(blocks.StructBlock):
    text = blocks.TextBlock()
    button_label = blocks.CharBlock(max_length=255)
    button_link = blocks.TextBlock()

    class Meta:
        template = 'blocks/cta.html'

    def get_speech_text(self, value):
        return "%s<break strength=\"x-strong\"/>%s" % (
            force_text(value.get('text')),
            force_text(value.get('button_label')))

class ExamplePage(SynthesizeSpeechMixin, Page):
    content = StreamField([
        ('example', ExampleBlock()),
    ])

    content_panels = Page.content_panels + [
        StreamFieldPanel('content'),
    ]

    def get_speech_text(self, request):
        from wagtailspeech.utils import get_speech_text_from_stream_field
        return get_speech_text_from_stream_field(request, self.main_content)

Settings

Available settings:

WAGTAIL_SPEECH_OUTPUT_FORMAT = 'mp3'
WAGTAIL_SPEECH_SAMPLE_RATE = '8000'
WAGTAIL_SPEECH_VOICE_ID = 'Joey'

For other values please read the documentation.

Tips

For proper pronunciations of your text and have more control over breaks, etc. we strongly advice you to read more about the supported tags of the Speech Synthesis Markup Language in Polly.

Roadmap

• Solution for limitations (max. 1500 chars / 5 min audio cut-off.)
• Add real-time streaming
• Celery support