DEV Community: Leanvox

Turn Any Article Into a Podcast Episode — No Mic Required

Leanvox — Tue, 24 Mar 2026 08:04:08 +0000

You wrote a great blog post. Maybe a deep research thread. A newsletter issue. A product explainer.

And then it sat there — read by a few hundred people, forgotten by most.

The problem isn't the content. It's the format. Audio reaches audiences that text never will. Commuters, gym-goers, people who scan rather than read. But turning written content into a podcast traditionally means: booking a studio, coordinating guests, recording multiple takes, and spending hours in editing software.

Or you paste a URL and click a button.

Meet Text-to-Podcast

LeanVox's Text-to-Podcast takes any article URL or raw text and converts it into a two-host podcast-style MP3 conversation — in seconds.

No equipment. No guests. No editing.

The two hosts — Alex (male) and Jordan (female) — have a natural back-and-forth dialogue covering your content. The result sounds like a real podcast episode, not a robot reading your article aloud.

How It Works

Script generation — Claude Haiku reads your content and writes a natural dialogue script between Alex and Jordan.
Audio rendering — The LeanVox Dialogue API renders the script using two distinct voices, producing a downloadable MP3.

Before the audio generates, you can view and edit the script directly in the tool. Toggle edit mode, make your changes, then hit generate.

It's also a live demo of what the LeanVox Dialogue API can do — if you're building a product that needs multi-voice narration, this is the capability you can tap into programmatically.

Choose Your Style

Style	Best For
Casual	Blog posts, newsletters, personal essays
Professional	Research papers, technical docs, industry reports
Debate	Opinion pieces, controversial topics, two-sided arguments

The Debate style is worth calling out. Instead of Alex and Jordan agreeing, they argue both sides of your topic. Great for op-eds, hot takes, or any content where tension makes it more listenable.

Who It's For

Newsletter writers — Repurpose every issue as a podcast episode. Double your reach, zero extra effort.
Bloggers — Give readers an audio option without starting a whole podcast.
Researchers & academics — Make dense papers more accessible.
Students — Convert study material into audio you can listen to on the move.
Developers — Explore the Dialogue API before building it into your own product.

Free to Use

👉 Try Text-to-Podcast at leanvox.com/tools/text-to-podcast

Want API access? Grab your keys from the LeanVox dashboard and check the docs.

More Free Tools

PDF-to-Audio — Convert PDFs to narrated audio.
Voice A/B Tester — Compare voices side-by-side before committing.

Your content is already written. Make it work harder.

Originally published at leanvox.com/blog

How to Pick the Right AI Voice Before You Commit to It

Leanvox — Mon, 23 Mar 2026 10:54:22 +0000

Here's a mistake developers and creators make constantly: they pick a voice by reading its name in a dropdown, generate a few thousand lines of audio, and then realize it sounds completely wrong for their project.

Too stiff. Too casual. Wrong energy for the brand.

Starting over is expensive — in time, money, and motivation.

LeanVox's Voice A/B Tester exists to stop that mistake before it happens. Pick two voices, enter your actual text, hear them both. Side by side. Before you build anything.

The Taste Test Before You Buy

Think of it like a wine flight, but for AI voices.

You wouldn't order a case of wine without tasting it. You shouldn't generate thousands of lines of audio without hearing the voice on your content first.

The A/B Tester is exactly that: a zero-commitment way to compare two voices on your actual script. Not a demo clip. Not a generic sample sentence. Your words, your tone, your use case.

How It Works

Pick Voice A and Voice B from the available voice library
Enter your text — use something representative of your actual project
Hit play — hear both rendered side by side
Pick the winner — then go build with confidence

Free to use. No account required.

Available Voices

Standard Voices

Voice	Character
Heart (Female)	Warm, natural, versatile
Bella (Female)	Clear, expressive, engaging
Michael (Male)	Confident, professional
Adam (Male)	Grounded, conversational

Pro Voices

Voice	Best For
Podcast Casual Male	Conversational shows, vlogs
Podcast Conversational Female	Interview-style content, storytelling
Narrator Warm Male	Audiobooks, long-form narration
Narrator Calm Female	Meditations, explainers, e-learning

Real Use Cases

Podcast producers — Should your show sound casual or polished? Run your intro script through Podcast Casual Male vs. Narrator Warm Male and the answer becomes obvious.

Audiobook narrators — Test on a full paragraph, not a single sentence — pacing and breath patterns matter at scale.

Game developers — Character voices are everything. Test multiple voices on the same line of dialogue before generating hundreds of lines of NPC audio.

Brand teams — Your brand voice is part of your brand. Hear it before you ship it.

App developers — Test voices against your actual UI copy before hardcoding a voice ID. Your users will thank you.

Why This Saves You Money

If you pick the wrong voice and re-generate everything, you've doubled the cost and the time. The A/B Tester is free. There's no good reason to skip the taste test.

Ready to Build After You Choose?

LeanVox API docs — integrate TTS directly into your app
Get an API key — free account, 200+ minutes, no credit card
PDF to Audio tool — convert documents to audio

Hear Before You Commit

Two minutes of testing now saves hours of regret later.

👉 Try the Voice A/B Tester free at leanvox.com/tools/voice-ab-tester

No login required. The right voice will be obvious.

Originally published at leanvox.com/blog

Convert Any PDF to Audio in Seconds — No Subscription Required

Leanvox — Mon, 23 Mar 2026 10:54:21 +0000

You've got a 40-page research paper to get through. Or a business book sitting in your downloads folder. Or a technical doc you keep putting off.

You don't always have time to sit down and read. You do have a commute, a gym session, or dishes to wash.

That's exactly what LeanVox's PDF to Audio tool is built for: drop in a PDF, get back an MP3. No subscription. No bloated desktop app. No friction.

How It Works

Upload your PDF — any PDF, any language
Pick a voice — choose from 238+ Standard and Pro voices
Generate — LeanVox extracts the text and runs it through the TTS API
Download your MP3 — ready to play anywhere

Language detection is automatic. If your PDF is in Spanish, French, or German, the tool handles it without any manual configuration.

Who This Is Actually For

Developers who want to audio-test documentation or onboarding content before shipping it
Students converting textbooks and research papers into commute-friendly audio
Creators and marketers turning newsletters or blog exports into audio content
Founders catching up on business books without carving out reading time
Anyone who reads slower than they listen

Free vs. Paid

Feature	Free (No Login)	Free Account	Pro
Pages per session	1	Unlimited	Unlimited
Standard voices	✅	✅	✅
Pro voices	❌	❌	✅
MP3 download	✅	✅	✅
Audio minutes included	—	200+	Per plan
Credit card required	No	No	No

Why Not Just Use Natural Reader or Speechify?

Tool	Starting Price	Pay-per-use	Voice count
Natural Reader	~$9.99/mo	❌	Limited
Speechify	~$9/mo	❌	Limited
LeanVox PDF to Audio	Free	✅ ~$0.003/page	238+

At roughly $0.003 per page (OCR + TTS combined), you're not locked into a monthly subscription you'll forget about.

Under the Hood (For the Curious)

The same pipeline is available via API:

PDF text extraction via OCR
LeanVox TTS API — the same API at leanvox.com/docs
MP3 output — streamed back or downloaded

Grab an API key at leanvox.com/dashboard/keys and run this programmatically — great for automating documentation audio, generating podcast scripts, or building audio-first products.

Try It Now — No Login Required

Your first page is free. No account, no credit card, no commitment.

👉 Convert your PDF to audio at leanvox.com/tools/pdf-to-audio

Create a free account and get 200+ minutes of audio to work with. Still no credit card.

Originally published at leanvox.com/blog

LeanVox is Now an Agent Skill: Add TTS to Claude Code, Cursor, and 37+ AI Agents in One Command

Leanvox — Fri, 13 Mar 2026 13:46:46 +0000

If you use an AI coding agent — Claude Code, Cursor, Windsurf, Copilot, OpenCode, or any of the 37+ agents on skills.sh — you can now give it the ability to generate speech, transcribe audio, and clone voices with a single command:

npx skills add leanvox/leanvox-skill

That's it. Your agent now knows how to use the LeanVox API.

What Is an Agent Skill?

Skills are reusable capability packages for AI agents. They give agents procedural knowledge — not just "what the API does" but how to use it correctly: which model tier to pick, when to use async, how to handle voice cloning, how to batch-generate dialogue.

The skills.sh registry is the npm for agent skills. Install once, and any compatible agent can immediately use those capabilities without you writing glue code or pasting in documentation.

What the LeanVox Skill Includes

Six helper scripts covering the full LeanVox API:

tts.sh — text-to-speech (sync for short text, async for long-form)
stt.sh — audio transcription with optional speaker diarization
dialogue.sh — multi-speaker conversation generation from a script
voiceover.sh — transcribe an audio file, edit the transcript, re-voice it
voices.sh — browse and search 238+ curated voices by category and gender
clone.sh — voice cloning from a reference audio file

Plus references/api-reference.md and references/voice-catalog.md — full endpoint docs and the complete voice catalog, so agents make informed decisions without hitting the docs site.

Pricing-Aware by Design

The skill teaches agents to pick the cheapest tier that works:

Tier	Cost	Best For
Standard	$0.005/1K chars	Fast narration, notifications, bulk generation
Pro	$0.01/1K chars	Expressive voices, podcasts, 238 curated voices
Max	$0.03/1K chars	Custom voice design from a text description

The skill defaults agents to Standard unless they specifically need Pro voices or Max-tier instruction-based voice design. A $0.03 Max call where Standard would have worked is money wasted — the skill prevents that.

What Your Agent Can Do Now

With the skill installed, you can tell your agent things like:

"Generate an MP3 of this blog post intro using a warm female narrator voice"
"Transcribe this meeting recording and give me a summary with speaker labels"
"Create a two-host dialogue between Alex and Jordan discussing this product launch"
"Clone my voice from voice-sample.wav and read this script"
"Find me a calm male narrator voice in the meditation category"
"Generate audio for all 500 dialogue lines in my game script"

The agent handles the API calls, model selection, async job management, and file downloads — you just describe what you want.

Works With 37+ Agents

The skill uses the SKILL.md format, compatible with Claude Code, Cursor, Windsurf, GitHub Copilot, OpenCode, Codex CLI, and 31+ more.

Install It

# Install the skill
npx skills add leanvox/leanvox-skill

# Set your API key
export LEANVOX_API_KEY="lv_live_..."

Get your API key (free signup credit — 200+ minutes of audio, voice cloning included) at leanvox.com/dashboard/keys.

The skill is open source: github.com/leanvox/leanvox-skill. PRs welcome.

Originally published at leanvox.com/blog

Automate Voice with LeanVox + n8n: Our Community Node is Live

Leanvox — Thu, 12 Mar 2026 14:38:02 +0000

If you use n8n for workflow automation, you can now add voice AI to any workflow. Our community node is live on npm.

Install it from Settings → Community Nodes in your n8n instance:

n8n-nodes-leanvox

That's it. No Docker config, no environment variables, no SDK installation.

What Can You Build?

The node covers the full LeanVox API — text-to-speech, speech-to-text, and multi-speaker dialogue. Here are some workflows that take minutes to set up:

Blog to Podcast

RSS feed triggers → extract article text → LeanVox Generate Speech → upload MP3 to S3 or your podcast host. Every new blog post automatically becomes an audio version.

Meeting Transcriber

Webhook receives recording → LeanVox Transcribe (with diarization + summary) → post summary to Slack. Know who said what without listening to the whole meeting.

Multilingual Voicemail

Form submission → LeanVox Generate Speech in 10 languages → email each version. One form, global reach.

Content Moderation Pipeline

Audio upload webhook → LeanVox Transcribe → scan transcript for flagged keywords → alert on Slack or email. Automate audio review at scale.

Available Operations

The node gives you 8 operations:

Generate Speech — text to audio using Standard (fast), Pro (238 curated voices), or Max (instruction-based voice design)
Generate Speech (Async) — for long text — kicks off a background job so your workflow doesn't time out
Check Job — poll an async job until complete
Dialogue — multi-speaker conversations with different voices per line
Transcribe — audio → text with optional speaker diarization and AI summary
List Voices — get all available voice IDs
List Curated Voices — browse 238 curated voices with preview audio
Check Balance — see your remaining credits

Setup in 60 Seconds

In n8n, go to Settings → Community Nodes → Install
Enter n8n-nodes-leanvox
Add a LeanVox API credential with your API key (get one here)
Drag the LeanVox node into any workflow

Example: Text to Speech in a Workflow

Add a Manual Trigger or Webhook node
Add the LeanVox node
Set Resource to Speech, Operation to Generate
Pick a model: standard for speed, pro for voice quality, max for custom voice instructions
Set a voice ID (e.g. podcast_conversational_female)
Pass your text

The node returns JSON with an audio_url you can pass to any downstream node — upload to S3, send via email, post to Slack, whatever your workflow needs.

Async for Long Content

For longer text (articles, chapters, scripts), use Generate Speech (Async). It queues a background job and returns a job_id. Chain it with the Check Job operation to poll until complete. No timeout issues, even for book-length content.

Pricing

Same credits as the API. Standard $0.005/1K chars · Pro $0.01/1K chars · Max $0.03/1K chars · Transcription $0.002/min. Full pricing.

Links

Originally published at leanvox.com/blog

Automate YouTube Shorts with AI Voice: Generate 50 Videos a Week with Python

Leanvox — Wed, 11 Mar 2026 10:56:44 +0000

YouTube Shorts rewards volume. Channels that post 5–10 times a day grow faster than those that post once a week.

Manual production at that volume is impossible. With an LLM writing scripts and a TTS API doing voiceover, you can generate 50 Shorts a week with a Python script that runs every morning.

The architecture

Script generation — Claude writes a 30–60 second script
Voiceover generation — LeanVox converts it to audio
Video assembly — ffmpeg combines voiceover with stock footage
Upload — YouTube Data API v3 schedules the video

Stage 1: Script generation

import anthropic

claude = anthropic.Anthropic()

def write_short_script(topic: str) -> str:
    response = claude.messages.create(
        model="claude-opus-4-5",
        max_tokens=300,
        messages=[{"role": "user", "content": f"""Write a YouTube Shorts script about: {topic}
Length: 45-60 seconds (~120-150 words). Start with a hook. End with a CTA. Return only the script."""}]
    )
    return response.content[0].text

Stage 2: Voiceover via CLI

For a single Short, use the CLI directly:

# Generate voiceover from script file
lvox generate \
  --model pro \
  --voice podcast_conversational_female \
  --speed 1.1 \
  --file script.txt \
  --output voiceover.mp3

Or via Python SDK with async jobs for batch processing:

from leanvox import Leanvox
import requests, os

client = Leanvox(api_key="lv_live_...")
CHANNEL_VOICE = "podcast_conversational_female"

def generate_voiceover(script: str, output_path: str) -> str:
    job = client.generate_async(
        text=script,
        model="pro",
        voice=CHANNEL_VOICE,
        speed=1.1,
    )
    result = job.wait()
    audio = requests.get(result.audio_url).content
    with open(output_path, "wb") as f:
        f.write(audio)
    return output_path

Stage 3: Video assembly

import subprocess, json

def get_audio_duration(audio_path: str) -> float:
    result = subprocess.run(
        ["ffprobe", "-v", "quiet", "-print_format", "json", "-show_streams", audio_path],
        capture_output=True, text=True
    )
    return float(json.loads(result.stdout)["streams"][0]["duration"])

def assemble_short(audio_path: str, background_video: str, output_path: str) -> str:
    duration = get_audio_duration(audio_path)
    subprocess.run([
        "ffmpeg", "-y", "-stream_loop", "-1",
        "-i", background_video, "-i", audio_path,
        "-t", str(duration), "-c:v", "libx264", "-c:a", "aac",
        "-vf", "scale=1080:1920",  # vertical 9:16 for Shorts
        output_path
    ], check=True, capture_output=True)
    return output_path

Clone your voice for brand consistency

with open("my_voice.wav", "rb") as f:
    voice = client.voices.clone(name="My Channel Voice", audio=f)
    client.voices.unlock(voice.voice_id)

job = client.generate_async(text=script, model="pro", voice=voice.voice_id, speed=1.1)

Or describe your channel persona with Max tier:

job = client.generate_async(
    text=script,
    model="max",
    instructions="Energetic tech educator, male, early 30s. Enthusiastic but not annoying. Clear and direct."
)

Weekly automated batch

import schedule, time

def weekly_batch():
    topics = get_this_weeks_topics()
    # Submit all jobs in parallel
    pending = [(topic, client.generate_async(text=write_short_script(topic), model="pro", voice=CHANNEL_VOICE)) for topic in topics]
    # Collect and assemble
    for topic, job in pending:
        result = job.wait()
        video = assemble_short(result.audio_url, "bg.mp4", f"queue/{hash(topic)}.mp4")
        schedule_youtube_upload(video, topic)

schedule.every().monday.at("06:00").do(weekly_batch)
while True:
    schedule.run_pending()
    time.sleep(60)

What it costs

A typical 45-second Short = ~750 characters.

Volume	Cost/video	Monthly
1 Short/day	$0.0075	$0.23
5 Shorts/day	$0.0075	$1.13
50 Shorts/day	$0.0075	$11.25

Your free signup credit covers 200+ minutes of audio — that's 130+ Shorts before you pay anything.

Try it

Browse voices · Get API key · Docs

Originally published at leanvox.com/blog

Self-Publishing an Audiobook with AI Voice: A Developer's Guide

Leanvox — Wed, 11 Mar 2026 10:56:16 +0000

ACX pays narrators $200–$400 per finished hour. A 10-hour audiobook costs $2,000–$4,000 with a professional narrator.

The same audiobook with AI narration: about $5.40.

A 10-hour audiobook is roughly 90,000 words — ~540,000 characters. At $0.01/1K chars (Pro tier), that's $5.40 total.

What works well with AI narration

Non-fiction — business books, self-help, technical guides
Educational content — courses, explainers, reference material
Newsletters and blogs — audio versions of written content
Short fiction — stories where a consistent narrator voice works

Choosing a narrator voice

LeanVox's narrator category has voices tuned for long-form listening. Preview before committing — there's a big difference between voices that sound great in a 10-second demo and ones that hold up over an hour.

For fiction with multiple characters, Max tier lets you describe specific voices:

CHARACTERS = {
    "detective_sarah": {
        "model": "max",
        "instructions": "Confident female detective, mid-30s. Direct speech. Occasionally dry humor. American accent."
    },
    "villain": {
        "model": "max",
        "instructions": "Cultured male villain, 50s. Smooth, controlled. Never raises voice. British accent."
    }
}

Processing a full book

You don't need to split your manuscript manually. LeanVox supports async jobs that accept full text files — the server handles chunking, processing, and reassembly automatically:

# Install the CLI
npm install -g leanvox

# Authenticate
lvox auth login

# Generate a full book from a .txt file — handles any length
lvox generate --use-async \
  --model pro \
  --voice narrator_warm_male \
  --file my_book.txt \
  --output audiobook.mp3

# Also works directly with .epub files
lvox generate --use-async \
  --model pro \
  --voice narrator_warm_male \
  --file my_book.epub \
  --output audiobook.mp3

The CLI submits an async job, polls for completion, and downloads the final file. A 90,000-word book typically completes in 10–20 minutes.

# Check job status
lvox jobs get <job-id>

# List all recent jobs
lvox jobs list

Via Python SDK

from leanvox import Leanvox
import requests

client = Leanvox(api_key="lv_live_...")

with open("my_book.txt") as f:
    manuscript = f.read()

# Submit async job — no chunking needed
job = client.generate_async(
    text=manuscript,
    model="pro",
    voice="narrator_warm_male",
)

result = job.wait()  # polls until complete, handles retries

audio = requests.get(result.audio_url).content
with open("audiobook.mp3", "wb") as f:
    f.write(audio)

print("📚 Audiobook complete!")

Normalize audio for distribution

ACX requires -16 LUFS:

ffmpeg -i audiobook.mp3 -af "loudnorm=I=-16:TP=-1.5:LRA=11" audiobook_normalized.mp3

What it costs

Book length	Word count	Chars	Pro tier cost
Novella (2h)	~20,000	~120K	$1.20
Short non-fiction (4h)	~40,000	~240K	$2.40
Full book (10h)	~90,000	~540K	$5.40

Your free signup credit covers 200+ minutes of audio — enough to produce a complete novella before spending a cent.

Distribution options

ACX / Audible — accepts MP3 at 192kbps+. AI-narrated books must be disclosed.
Findaway Voices — Apple Books, Kobo, libraries
Direct sale — Gumroad, Payhip, no disclosure requirement
Spotify / podcast RSS — publish chapter by chapter

No-code option: n8n

Prefer a visual workflow over Python? The LeanVox n8n community node lets you automate this without writing code. Install via Settings → Community Nodes in your n8n instance.

Try it

Browse narrator voices · Get your API key · Docs

Originally published at leanvox.com/blog

How Indie Devs Are Using AI Voice to Ship Games Faster (Without Hiring Voice Actors)

Leanvox — Wed, 11 Mar 2026 10:44:37 +0000

Voice acting is one of the last things indie devs ship and the first thing players notice is missing.

A professional voice actor costs $200–$400/hr. A full voiced RPG might need 50+ hours — a $10,000+ line item that kills the budget.

With a TTS API, you can voice every character for under $5 — and change any line instantly during development.

The workflow

Write dialogue in your dialogue system
Export as JSON or CSV
Run a generation script — each line gets an audio file
Import into Unity/Godot
Hook up to your trigger system

Once set up, changing a line takes 5 seconds and costs fractions of a cent.

Pick voices for your characters

LeanVox has 238+ voices organized by archetype:

Gaming — hero voices, villain voices, NPC archetypes
Narrator — world exposition, item descriptions
Meditation/Calm — sage characters, ancient beings
Kids — young player characters, sidekicks

Or use Max tier to describe any voice:

result = client.generate(
    text="The ancient prophecy speaks of one who will...",
    model="max",
    instructions="Elderly male wizard, deep gravelly voice, slow deliberate speech, hint of otherworldly echo"
)

Batch generate all dialogue (with async jobs)

Submit all lines in parallel using async jobs — much faster than waiting for each line sequentially:

import json
import os
import requests
from leanvox import Leanvox

client = Leanvox(api_key="lv_live_...")

with open("dialogue.json") as f:
    dialogue = json.load(f)

VOICES = {
    "town_guard": "gaming_gruff_male",
    "village_elder": "narrator_wise_elder",
    "young_hero": "podcast_casual_male",
    "mysterious_stranger": "narrator_dramatic_male",
    "merchant": "podcast_conversational_female",
}

os.makedirs("audio/dialogue", exist_ok=True)

# Submit all jobs in parallel
jobs = {}
for line in dialogue:
    output_path = f"audio/dialogue/{line['id']}.mp3"
    if os.path.exists(output_path):
        continue  # skip already generated

    voice = VOICES.get(line["character"], "podcast_casual_male")
    job = client.generate_async(text=line["text"], model="pro", voice=voice)
    jobs[job.job_id] = (line["id"], output_path)

print(f"Submitted {len(jobs)} jobs...")

# Collect results
for job_id, (line_id, output_path) in jobs.items():
    result = client.jobs.wait(job_id)
    audio = requests.get(result.audio_url).content
    with open(output_path, "wb") as f:
        f.write(audio)
    print(f"✅ {line_id}")

Or use the CLI for single lines:

lvox generate --model pro --voice gaming_gruff_male --file guard_line.txt --output guard_01.mp3

Unity integration

public class DialogueTrigger : MonoBehaviour
{
    [SerializeField] private AudioSource audioSource;

    public void PlayLine(string lineId)
    {
        StartCoroutine(LoadAndPlay(lineId));
    }

    private IEnumerator LoadAndPlay(string lineId)
    {
        var clip = Resources.Load<AudioClip>($"Audio/Dialogue/{lineId}");
        if (clip != null)
        {
            audioSource.clip = clip;
            audioSource.Play();
        }
        yield return null;
    }
}

Godot (GDScript)

extends Node
var http_request = HTTPRequest.new()

func generate_voice_line(text: String, voice: String) -> void:
    add_child(http_request)
    http_request.request_completed.connect(_on_request_completed)
    var headers = ["Authorization: Bearer " + OS.get_environment("LEANVOX_API_KEY"), "Content-Type: application/json"]
    var body = JSON.stringify({"text": text, "model": "pro", "voice": voice})
    http_request.request("https://api.leanvox.com/v1/tts/generate", headers, HTTPClient.METHOD_POST, body)

What does it cost?

Game type	Est. dialogue	Pro tier cost
Small game (200 lines)	~50K chars	$0.50
Mid-size RPG (2,000 lines)	~500K chars	$5.00
Large story-driven (5,000 lines)	~1.5M chars	$15.00

Your free signup credit covers 200+ minutes of audio — enough to voice a small game completely before spending a cent.

When to use a real voice actor

AI voice: all NPCs, ambient dialogue, narrator, tutorial prompts, procedural content.

Real voice actors: main protagonist, iconic villain, cutscene moments where delivery needs to be perfect.

Use both — AI voices everything, then swap key lines with pro recordings if your game gets traction.

No-code option: n8n

Prefer a visual workflow over Python? The LeanVox n8n community node lets you automate this without writing code. Install via Settings → Community Nodes in your n8n instance. Batch-generate voice lines from a spreadsheet — no scripting needed.

Try it

Browse the voice library · Get your API key · Docs

Originally published at leanvox.com/blog

AI Voice for Podcast Narration: Generate Professional Audio for $0.05/Episode

Leanvox — Wed, 11 Mar 2026 10:44:03 +0000

You don't need a recording booth, a microphone, or a consistent daily schedule to make a podcast anymore.

With a TTS API and a decent script, you can generate a 20-minute episode in under 60 seconds for about $0.05.

The use case

Solo narration podcasts — news roundups, book summaries, tech explainers, daily briefings
Multi-host formats — interview-style shows with two or more distinct voices
Branded audio content — consistent voice across hundreds of episodes
Translated episodes — same script, different language, same production cost

Picking a voice

LeanVox ships with 238+ pro voices organized by use case:

Podcast hosts — warm, conversational, sounds like someone you'd listen to for an hour
Narrators — authoritative, measured, great for explainer content
News anchors — crisp, professional, high credibility

Or clone your own voice. Record 30 seconds, upload it, and every episode sounds like you.

Single-host episode

from leanvox import Leanvox

client = Leanvox(api_key="lv_live_...")

result = client.generate(
    text=episode_script,
    model="pro",
    voice="podcast_conversational_female",
    speed=1.0,
)

print(result.audio_url)

Long episodes: use async jobs

For episodes longer than 10,000 characters, use the CLI with async jobs — no manual chunking needed:

# Generate from a script file — handles any length automatically
lvox generate --use-async \
  --model pro \
  --voice podcast_conversational_female \
  --file episode_script.txt \
  --output episode_042.mp3

Or via Python SDK:

import requests

job = client.generate_async(
    text=long_episode_script,
    model="pro",
    voice="podcast_conversational_female",
)

result = job.wait()  # polls until complete

audio = requests.get(result.audio_url).content
with open("episode_042.mp3", "wb") as f:
    f.write(audio)

Two-host conversation format

episode = client.dialogue(
    model="pro",
    gap_ms=500,
    lines=[
        {"text": "Welcome back to Syntax Error. Today we're talking about LLM inference costs.", "voice": "podcast_conversational_female", "language": "en"},
        {"text": "GPU costs have dropped 70% in two years and nobody's passing that on to developers.", "voice": "podcast_casual_male", "language": "en", "exaggeration": 0.6},
        {"text": "Strong take. Let's break down the numbers.", "voice": "podcast_conversational_female", "language": "en"},
    ],
)
print(episode.audio_url)

One call, one MP3, both voices.

Clone your voice for brand consistency

with open("my_voice_sample.wav", "rb") as f:
    voice = client.voices.clone(name="My Podcast Voice", audio=f)
    client.voices.unlock(voice.voice_id)

job = client.generate_async(text=episode_script, model="pro", voice=voice.voice_id)
result = job.wait()

Automate the full pipeline

import anthropic
from leanvox import Leanvox

claude = anthropic.Anthropic()
leanvox = Leanvox(api_key="lv_live_...")

def generate_daily_briefing(topic: str) -> str:
    response = claude.messages.create(
        model="claude-opus-4-5",
        max_tokens=1024,
        messages=[{"role": "user", "content": f"Write a 2-minute podcast script about: {topic}. Professional, conversational tone."}]
    )
    script = response.content[0].text

    job = leanvox.generate_async(text=script, model="pro", voice="podcast_conversational_female")
    result = job.wait()
    return result.audio_url

url = generate_daily_briefing("The latest in open-source AI models")

What does it cost?

For a 10-minute episode (~40,000 characters):

Tier	Rate	Episode cost	100 episodes/mo
Standard	$0.005/1K chars	$0.20	$20
Pro (with cloning)	$0.01/1K chars	$0.40	$40

A voice actor charges $200–$500/hr. A 10-minute episode = 30–60 min studio time.

No-code option: n8n

Prefer a visual workflow over Python? The LeanVox n8n community node lets you automate this without writing code. Install via Settings → Community Nodes in your n8n instance. Build a full podcast pipeline — RSS → narration → upload. Zero code.

Try it

Browse the voice library · Get your API key · Docs

Originally published at leanvox.com/blog

Voice-Over Studio: Re-voice Any Audio for $0.06 (16 Cheaper Than ElevenLabs)

Leanvox — Fri, 06 Mar 2026 14:51:26 +0000

ElevenLabs charges $1.00+ to re-voice a 5-minute audio clip.

We built a better workflow for $0.06.

What is Voice-Over Studio?

Upload any audio. We transcribe it, detect every speaker, and let you assign a different voice to each one. Then we re-generate the whole thing with our TTS models.

Change the host's voice. Replace the guest. Edit the script. Remove filler words. One workflow, four steps.

How it works

1. Upload your audio

Drop an mp3, wav, m4a, or any common format. We transcribe it automatically using Whisper V3 with speaker diarization — each speaker gets their own labeled track.

2. Edit the transcript

Fix typos, rewrite lines, remove filler words. Each segment is a plain text field. The original audio stays playable for reference.

3. Pick a voice for each speaker

Browse 238+ curated voices by category: podcast hosts, narrators, gaming characters, meditation guides, news anchors, kids. Preview any voice before selecting.

4. Generate and download

Every segment gets generated with the assigned voice, then stitched together. One audio file, ready to download.

The economics

Provider	5-min re-voicing	Multi-speaker
ElevenLabs Dubbing	~$1.00	✅
LeanVox Voice-Over	~$0.06	✅

16× cheaper. Your free signup credit covers 200+ minutes of audio — that's 15+ sessions.

Breakdown:

Transcription: $0.002/min × 5 min = $0.01
TTS generation (Pro): $0.01/1K chars × ~5,000 chars = $0.05
Total: ~$0.06

Via API

from leanvox import Leanvox

client = Leanvox(api_key="lv_live_...")

# Step 1: Transcribe + detect speakers
transcript = client.audio.transcribe(
    file="podcast.mp3",
    features=["transcribe", "diarize"]
)

# Step 2: Assign voices and re-generate
segments = [
    {
        "speaker": "host",
        "text": seg.text,
        "voice": "podcast_casual_host_m"
    }
    for seg in transcript.segments
    if seg.speaker == "SPEAKER_0"
]

result = client.tts.dialogue(segments=segments, model="pro")

Who is this for?

Podcast producers — swap a guest voice, fix a bad recording session
Content creators — A/B test different host voices
Localization teams — re-voice content for different markets
Accessibility — convert any audio to a clearer, more articulate voice
Developers — automate voice-over pipelines via API

Try it

Open Voice-Over Studio — live in your dashboard now. No new account, no separate billing. Your existing API key and $1 signup credit work.

Get API key · Docs · leanvox.com

Audio Intelligence: Transcription + Speaker Labels for $0.002/min (with Free Diarization)

Leanvox — Wed, 04 Mar 2026 10:00:11 +0000

LeanVox started as a text-to-speech API. Today it handles both sides of audio.

Meet Audio Intelligence — transcription, speaker diarization, and AI summarization in a single API call. Same API key. Same dashboard. No new account.

One endpoint. Three outputs.

from leanvox import Leanvox

client = Leanvox(api_key="lv_live_...")

result = client.audio.transcribe(
    file="meeting.mp3",
    features=["transcribe", "diarize", "summarize"]
)

print(result.formatted_transcript)
# SPEAKER_0: Welcome to the show.
# SPEAKER_1: Thanks for having me.

print(result.summary)
# "Team discussed Q1 roadmap priorities..."

Or with Node.js:

const result = await client.audio.transcribe({
  file: "meeting.mp3",
  features: ["transcribe", "diarize", "summarize"]
})

console.log(result.formatted_transcript)
console.log(result.summary)

Pricing that actually makes sense

We benchmarked Whisper Large V3 + Pyannote 3.1 on dedicated GPU hardware:

Feature	LeanVox	AssemblyAI	Deepgram
Transcription	$0.002/min	$0.0025/min	$0.0043/min
Speaker diarization	Free	+$0.007/min	+$0.014/min
Total (transcript + speakers)	$0.002/min	$0.0095/min	$0.018/min

4.75× cheaper than AssemblyAI. 9× cheaper than Deepgram. Speaker labels included free.

Your free signup credit covers 200+ minutes of TTS audio and 500 minutes of transcription. A 1-hour meeting costs $0.12.

Why diarization is free

Most providers charge extra for speaker detection. We don't — our infrastructure makes it nearly zero marginal cost (adds <0.5s to processing). We'd rather bundle it and give you a better product.

Works with the MCP server too

No code required with Claude:

{
  "mcpServers": {
    "leanvox": {
      "command": "npx",
      "args": ["leanvox-mcp"],
      "env": { "LEANVOX_API_KEY": "lv_live_..." }
    }
  }
}

Tell Claude: "Transcribe this audio file and give me a summary with speaker labels." Zero code.

What's supported

Formats: mp3, wav, ogg, flac, m4a, webm (up to 500MB)
Languages: 99 (auto-detected or specify)
Processing: Sync for files ≤5 min, async with webhook callbacks for longer
SDKs: Python and Node.js (v0.3.0)

Getting started

# Install
pip install leanvox  # or npm install leanvox

# Transcribe
curl -X POST https://api.leanvox.com/v1/audio/transcribe \
  -H "Authorization: Bearer lv_your_key_here" \
  -F "file=@audio.mp3"

No-code option: n8n

Prefer a visual workflow? The LeanVox n8n community node includes a Transcribe operation — build no-code audio pipelines (webhook → transcribe → Slack summary) in minutes. Install via Settings → Community Nodes in n8n.

→ Quickstart guide · API reference · Get your API key

Voice Remix: Start from Any Voice, Make It Yours

Leanvox — Mon, 02 Mar 2026 15:56:17 +0000

You found a voice you like. But it's not quite right.

Maybe the narrator is perfect but you want them younger. The podcast host has the right energy but you need a British accent. The meditation guide is calming but you want it deeper.

Now you can fix that in one click.

Introducing Voice Remix

Every curated Pro voice now has a Remix button. Click it, and you jump to Max tier with the voice's design prompt pre-loaded — ready to tweak.

The voice narrator_warm_male was created with:

"A warm, resonant male narrator in his late 40s with a deep baritone voice and measured, confident pacing"

Hit Remix, and that prompt appears in the Max editor. Change "late 40s" to "early 20s." Swap "baritone" for "tenor." Add "with a slight Irish accent." Generate.

You just designed a custom voice in 10 seconds, starting from something you already liked.

Start from a Voice

Don't know where to begin? The Max tier now has a "Start from a voice" dropdown — 200+ curated voice prompts organized by category:

🎙️ Podcast hosts
📖 Narrators
🧘 Meditation guides
🎮 Gaming characters
👶 Kids content
📰 News anchors

There's even a 🎲 Surprise Me button that picks a random voice — perfect for exploring.

How It Works

In the Voice Studio:

Browse voices at leanvox.com/voices
Find one you like → click 🎵 Remix
Modify the prompt → Generate

Via API:

# Get curated voices with their design prompts
voices = client.voices.list(include_prompt=True)

# Start from a preset, modify it
prompt = voices[0].voice_instructions
custom = prompt.replace("measured pacing", "fast and energetic")

result = client.generate(
    text="Breaking news from the studio...",
    model="max",
    voice_instructions=custom
)

Why This Matters

Most TTS platforms give you a fixed catalog. Pick a voice, take it or leave it.

LeanVox gives you a starting point and a paintbrush. Every curated voice is a prompt you can remix into something new. The catalog isn't a ceiling — it's a floor.

→ Try Voice Remix · Get your API key · Docs