fix(providers): use native ElevenLabs API on desktop to avoid unspeech proxy 401 by Hanfeng-Lin · Pull Request #1107 · moeru-ai/airi

Hanfeng-Lin · 2026-03-04T11:56:04Z

This is a follow-up to the previous PR #1100 , addressing the rejection reason.

Problem

The previous PR called ElevenLabs API directly from all platforms, which broke the web version due to CORS — browsers block cross-origin requests to api.elevenlabs.io since it doesn't return Access-Control-Allow-Origin headers.

The underlying issue remains: the public unspeech.hyp3r.link proxy causes HTTP 401 errors for Free Tier users because many users share the same proxy IP, triggering ElevenLabs' abuse detection.

Solution

Use isStageTamagotchi() to branch by platform:

Desktop: Electron's renderer is not subject to browser CORS rules, so direct cross-origin fetch to api.elevenlabs.io works, avoiding the shared-IP 401 issue.
Web: Keeps routing through the configured unspeech proxy — required for CORS compliance.

…h proxy 401

gemini-code-assist · 2026-03-04T11:56:32Z

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request resolves persistent HTTP 401 errors for ElevenLabs Free Tier users by intelligently routing API requests based on the application platform. Desktop clients now directly access the native ElevenLabs API, circumventing issues with shared proxy IPs and CORS limitations. Web clients continue to utilize the unspeech proxy to ensure CORS compliance. This change significantly improves the reliability of ElevenLabs integration for desktop users while maintaining functionality for web users.

Highlights

ElevenLabs API Routing: Implemented platform-specific routing for ElevenLabs API calls, using the native API for desktop (Electron) and the unspeech proxy for web.
401 Error Resolution: Addressed the HTTP 401 errors experienced by ElevenLabs Free Tier users on desktop by bypassing the shared-IP unspeech proxy.
Custom Fetch Logic: Introduced custom fetch logic for desktop to transform OpenAI-format requests into ElevenLabs' native API format.
Voice Listing Enhancement: Updated the voice listing mechanism to fetch voices directly from ElevenLabs for desktop and via the unspeech proxy for web, while maintaining consistent voice reordering.

Changelog

packages/stage-ui/src/stores/providers.ts
- Modified the baseUrl for ElevenLabs to dynamically switch between the native ElevenLabs API and the unspeech proxy based on whether the application is running on desktop (isStageTamagotchi()).
- Refactored the createProvider function for ElevenLabs to include a custom fetch implementation for desktop, translating OpenAI-format requests to ElevenLabs' native text-to-speech endpoint.
- Updated the listVoices capability to perform platform-specific voice fetching: direct API calls for desktop and proxy-based calls for web, followed by a unified voice reordering logic.

Activity

No human activity has been recorded on this pull request since its creation.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request introduces platform-specific logic for the ElevenLabs provider to use the native API on desktop, avoiding CORS issues and potential 401 errors from the public proxy. The changes are well-structured and address the problem described. I've identified a few areas for improvement regarding error handling, cleaner JSON payload construction, and a potential bug in array manipulation. My suggestions aim to enhance the robustness and maintainability of the new implementation.

_{Note: Security Review did not run due to the size of the PR.}

gemini-code-assist · 2026-03-04T12:01:05Z

+                  voice?: string
+                  model?: string
+                }
+                const voiceId = encodeURIComponent(body.voice ?? '')


If body.voice is undefined or an empty string, voiceId will be empty. This will likely cause the fetch call on line 899 to fail with a 404 error, as the URL will be .../text-to-speech/. It would be more robust to add a check for an empty voiceId and return an appropriate error response before making the API call.

gemini-code-assist · 2026-03-04T12:01:05Z

+          const lo = Math.min(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : voices.length - 1 }))
+          const hi = Math.max(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : 0 }))
+          return [...voices.slice(0, lo), ...voices.slice(hi + 1), ...voices.slice(lo, hi + 1)]


This logic for rearranging voices has a bug when neither 'Aria' nor 'Bill' are found in the voices array. In that case, lo becomes voices.length - 1 and hi becomes 0, causing lo > hi. The spread syntax on the next line will then duplicate parts of the array. A more robust implementation would handle this edge case explicitly.

Suggested change

const lo = Math.min(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : voices.length - 1 }))

const hi = Math.max(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : 0 }))

return [...voices.slice(0, lo), ...voices.slice(hi + 1), ...voices.slice(lo, hi + 1)]

const ariaIdx = voices.findIndex(v => v.name.includes('Aria'));

const billIdx = voices.findIndex(v => v.name.includes('Bill'));

if (ariaIdx === -1 && billIdx === -1) {

return voices;

}

const lo = Math.min(ariaIdx > -1 ? ariaIdx : Infinity, billIdx > -1 ? billIdx : Infinity);

const hi = Math.max(ariaIdx, billIdx);

return [...voices.slice(0, lo), ...voices.slice(hi + 1), ...voices.slice(lo, hi + 1)];

gemini-code-assist · 2026-03-04T12:01:05Z

+                    voice_settings: {
+                      stability: voiceSettings.stability ?? 0.5,
+                      similarity_boost: voiceSettings.similarityBoost ?? 0.75,
+                      style: voiceSettings.style,
+                      use_speaker_boost: voiceSettings.useSpeakerBoost,
+                    },


If voiceSettings.style or voiceSettings.useSpeakerBoost are undefined, they will be stringified as null in the JSON payload. While many APIs handle null gracefully, it's safer and cleaner to omit these properties from the payload if they are not defined. You can use conditional spreading to achieve this.

Suggested change

voice_settings: {

stability: voiceSettings.stability ?? 0.5,

similarity_boost: voiceSettings.similarityBoost ?? 0.75,

style: voiceSettings.style,

use_speaker_boost: voiceSettings.useSpeakerBoost,

},

voice_settings: {

stability: voiceSettings.stability ?? 0.5,

similarity_boost: voiceSettings.similarityBoost ?? 0.75,

...(voiceSettings.style !== undefined && { style: voiceSettings.style }),

...(voiceSettings.useSpeakerBoost !== undefined && { use_speaker_boost: voiceSettings.useSpeakerBoost })

}

nekomeowww · 2026-03-04T15:22:52Z

But the request still happens right inside of the browser - provided by Electrom, which is based on Chromium, the CORS issue still exists...

Hanfeng-Lin · 2026-03-04T19:09:36Z

The desktop version has already been tested and confirmed working without CORS issue. The network requests reach api.elevenlabs.io and receive valid responses, which would be impossible if browser-level CORS were enforced. The web end was also tested and the proxy is working as normal.
I think Electron has more relaxed security policies which could bypass the CORS restrictions, either by webPreferences: { webSecurity: false } in BrowserWindow or session.webRequest.onHeadersReceived to inject CORS headers into responses

shinohara-rin · 2026-03-06T21:15:31Z

$ curl -sI -X OPTIONS https://api.elevenlabs.io/v1/text-to-speech/test
-H "Origin: http://localhost"
-H "Access-Control-Request-Method: POST"

access-control-allow-origin: *
access-control-allow-headers: *

strange things, looks like elevenlabs return CORS headers as of now(or is it just me???)

…ts CORS

Hanfeng-Lin · 2026-03-07T02:26:09Z

Thanks @shinohara-rin for pointing out! I've updated the branch to bypass the unspeech proxy and call ElevenLabs' native API directly across all platforms. Previously this was Desktop-only due to CORS issues, but ElevenLabs now returns 'Access-Control-Allow-Origin: *'.

Both platforms were tested successfully

shinohara-rin

It looks a bit dirty to me to handcraft an http client here, this might only work as a temporary fix.

shall we consider supporting the elevenlabs native api? @nekomeowww
with CORS out of the way, we still have the anti-abuse problem to deal with

Hanfeng-Lin · 2026-03-07T10:40:41Z

If we're sticking with the native ElevenLabs API (because this solves the 401 redirect issue caused by shared IPs once and for all), I can extract the dirty work related to fetching and create a dedicated adapter file in src/stores/providers/elevenlabs/native.ts, exposing a clean createNativeElevenLabsProvider(). This way, providers.ts can now function like before, simply calling a single line of code.

…providers clean

github-actions · 2026-03-07T14:06:33Z

⏳ Approval required for deploying to Cloudflare Workers (Preview) for stage-web.

Name	Link
🔭 Waiting for approval	For maintainers, approve here

Hey, @nekomeowww, @sumimakito, @luoling8192, @LemonNekoGH, kindly take some time to review and approve this deployment when you are available. Thank you! 🙏

…, moeru-ai#1280, moeru-ai#1107, moeru-ai#1222, moeru-ai#1065)

fix(providers): use native ElevenLabs API on desktop to avoid unspeec…

16806fc

…h proxy 401

gemini-code-assist Bot reviewed Mar 4, 2026

View reviewed changes

refactor(providers): use native API directly as ElevenLabs now suppor…

742bc0b

…ts CORS

shinohara-rin reviewed Mar 7, 2026

View reviewed changes

refactor(providers): isolate elevenlabs native fetch adapter to keep …

727a922

…providers clean

Hanfeng-Lin force-pushed the elevenlabs_debug branch from e747f73 to 727a922 Compare March 7, 2026 11:24

Merge branch 'main' into elevenlabs_debug

d8fc14c

dasilva333 added a commit to dasilva333/airi that referenced this pull request Mar 13, 2026

feat: integrate 6 critical upstream fixes (moeru-ai#1124, moeru-ai#1312…

e5ae6ae

…, moeru-ai#1280, moeru-ai#1107, moeru-ai#1222, moeru-ai#1065)

dasilva333 added a commit to dasilva333/airi that referenced this pull request Mar 14, 2026

feat: integrate 6 critical upstream fixes (moeru-ai#1124, moeru-ai#1312…

54f4780

…, moeru-ai#1280, moeru-ai#1107, moeru-ai#1222, moeru-ai#1065)

nekomeowww added scope/providers Scope related to providers we support bug/providers Some providers aren't working pr-review/hold/unsure Pull Request that unsure about purpose, not sure if needed labels Mar 26, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(providers): use native ElevenLabs API on desktop to avoid unspeech proxy 401#1107

fix(providers): use native ElevenLabs API on desktop to avoid unspeech proxy 401#1107
Hanfeng-Lin wants to merge 4 commits intomoeru-ai:mainfrom
Hanfeng-Lin:elevenlabs_debug

Hanfeng-Lin commented Mar 4, 2026

Uh oh!

gemini-code-assist Bot commented Mar 4, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Uh oh!

nekomeowww commented Mar 4, 2026

Uh oh!

Hanfeng-Lin commented Mar 4, 2026

Uh oh!

shinohara-rin commented Mar 6, 2026

Uh oh!

Hanfeng-Lin commented Mar 7, 2026

Uh oh!

shinohara-rin left a comment

Uh oh!

Hanfeng-Lin commented Mar 7, 2026

Uh oh!

github-actions Bot commented Mar 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-          const lo = Math.min(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : voices.length - 1 }))
-          const hi = Math.max(...['Aria', 'Bill'].map((n) => { const i = voices.findIndex(v => v.name.includes(n)); return i !== -1 ? i : 0 }))
-          return [...voices.slice(0, lo), ...voices.slice(hi + 1), ...voices.slice(lo, hi + 1)]
+          const ariaIdx = voices.findIndex(v => v.name.includes('Aria'));
+          const billIdx = voices.findIndex(v => v.name.includes('Bill'));
+          if (ariaIdx === -1 && billIdx === -1) {
+            return voices;
+          }
+          const lo = Math.min(ariaIdx > -1 ? ariaIdx : Infinity, billIdx > -1 ? billIdx : Infinity);
+          const hi = Math.max(ariaIdx, billIdx);
+          return [...voices.slice(0, lo), ...voices.slice(hi + 1), ...voices.slice(lo, hi + 1)];

Uh oh!

Conversation

Hanfeng-Lin commented Mar 4, 2026

Problem

Solution

Uh oh!

gemini-code-assist Bot commented Mar 4, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

nekomeowww commented Mar 4, 2026

Uh oh!

Hanfeng-Lin commented Mar 4, 2026

Uh oh!

shinohara-rin commented Mar 6, 2026

Uh oh!

Hanfeng-Lin commented Mar 7, 2026

Uh oh!

shinohara-rin left a comment

Choose a reason for hiding this comment

Uh oh!

Hanfeng-Lin commented Mar 7, 2026

Uh oh!

github-actions Bot commented Mar 7, 2026

⏳ Approval required for deploying to Cloudflare Workers (Preview) for stage-web.

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants