/voice requires openai quota. I have with openrouter but in deed, it can be done offline

[jechaviz]

Issue

I got this message using my openrouter conf

Recording, press ENTER when done... 6.6sec ██████████
Unable to transcribe C:\Users\jesus\AppData\Local\Temp\tmph7mnknah.wav: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}
But I'm not using openai but openrouter.

---

In fact we can use speech to text like in this article
https://medium.com/@nimritakoul01/offline-speech-to-text-in-python-f5d6454ecd02

---

or using speech to text realtime with this opensource library
https://github.com/KoljaB/RealtimeSTT

---

or using whisper cpp implementation offline:
https://github.com/ggerganov/whisper.cpp
(it quotes python bindings in Readme here)
ggerganov/whisper.cpp#9
I'm testing this:
https://github.com/abdeladim-s/pywhispercpp

---

Also found this:
https://github.com/chidiwilliams/buzz

---

We can install whisper from openai offline (free) with this
https://scoop.sh/#/apps?q=whisper (for windows users)
and download model cloning
https://github.com/ggerganov/whisper.cpp
and running download-ggml-model as in documentation.

---

also look a this list for other alternatives, including the article creator edenai.co:
https://www.edenai.co/post/top-free-speech-to-text-tools-apis-and-open-source-models
edenai.co is an alternative for openrouter but includes more providers to include:
Generative Al
Image
Text
Speech
OCR & Document Parsing
Translation

---

Thanks.

Version and model info

Aider v0.59.1
Main model: openrouter/anthropic/claude-3.5-sonnet with diff edit format, infinite output
Weak model: openrouter/anthropic/claude-3-haiku-20240307
Git repo: .git with 1 files
Repo-map: using 1024 tokens, auto refresh

[curran]

I arrived at this issue after seeing the error:

Unable to transcribe /tmp/tmpnmerv9fq.wav: Error code: 405

Nowhere in the instructions does it say that voice mode requires you to configure an OpenAI API key.

Places I would have expected this:

• https://aider.chat/docs/usage/voice.html#how-to-use-voice-to-code
• https://aider.chat/docs/install/optional.html#enable-voice-coding
• https://aider.chat/docs/config/options.html#voice-settings
• https://aider.chat/docs/config/aider_conf.html

I'm also using OpenRouter, and have not configured the OpenAI key. I'm guessing that I need to set up an actual OpenAI API Key, which is a bit tricky since I already configured Aider like this:

openai-api-key: <OpenRouter API Key>
openai-api-base: https://openrouter.ai/api/v1

What's the best way to proceed?

[curran]

Judging by this

https://aider.chat/docs/llms/openrouter.html

it appears I must at this point add configuration to my system that lives outside of the Aider YML Conf. I wish I could configure the OpenRouter key from the YML.