Narrate Your Blog Posts Without Paying $330/Month

Why your blog needs an audio version

The way people consume content has fundamentally shifted. According to Edison Research's Infinite Dial 2024 report, podcast listenership surpassed 500 million globally in 2024. Over 60% of Americans listen to audio content weekly. And platforms like Medium, Substack, and WordPress have all added audio players because readers are asking for them.

Adding audio to your blog isn't a gimmick. It's a multiplier. The same article you already wrote can now reach people who are commuting, exercising, cooking, or simply prefer listening over reading. You're not creating new content. You're unlocking the content you've already made.

Audio also improves accessibility. Visually impaired readers, people with dyslexia, and non-native speakers all benefit from having a spoken version of your writing. Screen readers exist, but they sound robotic. A natural-sounding narration makes your content genuinely pleasant to listen to.

The $4,000-a-year problem

So you're convinced. Audio versions are worth it. You start looking into narration services. And that's where the sticker shock hits.

ElevenLabs charges $99–$330/month for their professional tiers. Google Cloud TTS bills per character, and at scale that adds up to hundreds per month. Amazon Polly, Azure Speech, PlayHT: all metered, all cloud, all requiring you to upload every word you write to someone else's servers.

George Pu, who runs the publication Founder Reality, almost signed up for ElevenLabs at $330/month. Then he tried running an open-source model on his own laptop. "Sounds fine," he wrote. "200 posts a month. Costs me electricity." He'd nearly paid $4,000 a year to rent a model he could run himself.

For individual bloggers, newsletter writers, and small publishers, these prices are hard to justify. You're already doing the work of writing, so why should turning that text into speech cost more than your hosting?

The DIY approach: powerful but complex

George didn't just identify the problem. He built a solution. He created "Ghost Narrator," a fully self-hosted pipeline that automatically narrates every new blog post. Here's the flow he shared:

No API keys. No subscriptions. No monthly bill. The whole thing runs on hardware he already owns. It's a genuinely impressive setup, and proof that the technology for local narration is mature.

But here's the catch: building Ghost Narrator required stitching together multiple open-source tools, writing custom automation scripts, managing Python environments and model versions, and being comfortable debugging things when they break. George is a technical founder, and this is his comfort zone. For most bloggers, course creators, and content marketers, this pipeline is aspirational but impractical.

What if you didn't have to build it?

Murmur gives you the same result: on-device narration with zero recurring costs, packaged in a native Mac app. No terminal, no Python scripts, no model downloads, no glue code. You open the app, paste your text, and click generate.

Under the hood, Murmur runs the Kokoro TTS model (82 million parameters) using MLX, Apple's machine learning framework optimized for Apple Silicon. The model runs entirely on your chip, and your text never touches the internet.

Both approaches share the same philosophy: your content stays on your machine, you pay nothing per generation, and you're not locked into any subscription. The difference is who they're built for.

How to narrate blog posts with Murmur

Here's the step-by-step workflow. From opening the app to having an embeddable audio file, the whole process takes about a minute.

Why running locally changes everything

The shift from cloud to local isn't just about saving money, though that's a big part of it. Running TTS on your own machine changes the relationship between you and your tools.

Batch narrate your entire archive

Most blogs have a back catalog: dozens or hundreds of posts that would benefit from audio but were never narrated. Going through them one by one would take forever.

Murmur's batch queue solves this. Import multiple articles, assign a voice to each (or use the same voice for all), and let them process in sequence. Walk away, come back, and find a folder of finished audio files. Some users narrate their entire archive over a single weekend.

The SEO case for audio blog posts

Search engines reward pages that keep visitors engaged. An audio player on your blog post increases time-on-page, reduces bounce rate, and signals to Google that your content is worth staying for.

Audio also opens distribution channels that text alone can't reach. Repurpose your narrated posts as podcast episodes, share audio clips on social media, or submit them to audio aggregators. Each channel drives backlinks and traffic to the original post.

And with Google increasingly surfacing multimedia results, having an audio version of your content gives you an edge over competitors who only publish text. It's a small effort for a compounding advantage.

Built for people who write

Murmur isn't a developer tool or an API wrapper. It's a Mac app designed for people who create written content and want to add audio without changing their workflow.

Frequently asked questions