Migrate from ElevenLabs to Railwail

TL;DR — Switch in Under 15 Minutes

• Drop the elevenlabs SDK — use the OpenAI SDK's audio.speech API
• ElevenLabs voice IDs are mapped to Railwail voice slugs (Rachel → eleven-rachel, etc.)
• Multilingual v3, Turbo v2.5, Flash v2 models all supported
• EU-hosted endpoint, EUR billing
• Plus access to OpenAI TTS, Bark, and 275+ other models on the same key

Why Move Off ElevenLabs?

ElevenLabs is the gold standard for realistic TTS. The trade-offs for EU teams: USD-only billing, US-default routing, no EU residency guarantee on the free / starter tiers, and a proprietary SDK that does not match the rest of your AI stack. Railwail exposes ElevenLabs voices through the OpenAI-compatible audio.speech API, with EU hosting and EUR billing.

Step 1 — Get a Railwail API Key

Sign up at railwail.com and generate a key.

Step 2 — Voice ID Mapping

ElevenLabs uses 20-character voice IDs like 21m00Tcm4TlvDq8ikWAM. Railwail uses human-readable slugs prefixed with eleven-. All public ElevenLabs voices have a Railwail equivalent.

For custom voices (Voice Lab clones), pass the original ElevenLabs voice ID directly — Railwail recognises 20-char IDs and routes them through to your ElevenLabs Voice Lab account (requires you to link the account in Railwail dashboard).

Step 3 — Replace the SDK Call

TypeScript / JavaScript

Before (ElevenLabs):

import { ElevenLabsClient } from "elevenlabs";

const el = new ElevenLabsClient({ apiKey: process.env.ELEVENLABS_API_KEY });

const audio = await el.generate({
  voice: "Rachel",
  text: "Hello world",
  model_id: "eleven_multilingual_v3",
});
// audio is a Readable stream

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.RAILWAIL_API_KEY,
  baseURL: "https://api.railwail.com/v1",
});

const res = await client.audio.speech.create({
  model: "eleven-multilingual-v3",
  voice: "eleven-rachel",
  input: "Hello world",
  response_format: "mp3",
});
const buffer = Buffer.from(await res.arrayBuffer());
require("fs").writeFileSync("out.mp3", buffer);

Python

from openai import OpenAI

client = OpenAI(
    api_key=os.environ["RAILWAIL_API_KEY"],
    base_url="https://api.railwail.com/v1",
)

with client.audio.speech.with_streaming_response.create(
    model="eleven-multilingual-v3",
    voice="eleven-rachel",
    input="Hello world",
    response_format="mp3",
) as resp:
    resp.stream_to_file("out.mp3")

cURL

curl https://api.railwail.com/v1/audio/speech \
  -H "Authorization: Bearer $RAILWAIL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "eleven-multilingual-v3",
    "voice": "eleven-rachel",
    "input": "Hello world",
    "response_format": "mp3"
  }' \
  -o out.mp3

API Endpoint Mapping

Model Mapping

Pricing Comparison (per 1k characters, May 2026)

Why Railwail Over ElevenLabs Direct

• EU billing in EUR with VAT receipts
• Same ElevenLabs voices at same prices
• OpenAI-compatible audio.speech API — unify with your text/image/STT stack
• Plus alternative TTS providers (OpenAI TTS, Bark, Coqui XTTS) on the same key
• Built-in playground at railwail.com/models for voice preview
• Per-key character quotas

FAQ

Can I use my custom Voice Lab clones?

Yes. Link your ElevenLabs account in the Railwail dashboard (Settings → Linked Accounts → ElevenLabs), then pass your custom voice_id (20-char string) instead of a Railwail slug. Cost is billed by ElevenLabs through Railwail at cost-pass-through pricing.

Are stability and similarity_boost controls preserved?

Yes. Pass voice_settings: { stability: 0.5, similarity_boost: 0.75, style: 0.0, use_speaker_boost: true } as an extension parameter.

What audio formats are returned?

response_format supports mp3 (default), pcm_16000, pcm_22050, pcm_24000, pcm_44100, ulaw_8000, opus.

Is streaming supported?

Yes. Pass stream: true to receive audio chunks as they generate, which is critical for real-time apps like voice agents.

What about ElevenLabs Conversational AI / agents?

ElevenLabs' Conversational AI is a full agent product, not exposed via Railwail. For equivalent functionality, build a custom agent using Railwail's chat.completions + audio.speech + audio.transcriptions endpoints.

Is dubbing / audio-to-audio supported?

ElevenLabs Dubbing is a separate product. For dubbing workflows, transcribe with Railwail's audio.transcriptions, translate with chat.completions, then re-synthesise with audio.speech.

Next Steps

• Sign up at railwail.com
• Generate an API key
• Replace elevenlabs SDK with OpenAI SDK
• Map voice IDs to Railwail slugs (eleven-rachel etc.)
• Read the reference at railwail.com/docs
• Browse voices at railwail.com/models?category=speech_tts
• Compare pricing at railwail.com/pricing