real time. with out wait whole reply done, button actually appear.

[happyf-weallareeuropean]

what is this challenge of this feature?

[adamlui]

@happyf-weallareeuropean currently the script just relies on the button (front-end method) so for your feature requires back-end method + API allows speak before completion but I never worked w/ this method before but I can look into it

[adamlui]

I just played with the API for awhile and it was too hardcore to implement from IDK what to do w/ the response. But this is where I got (xhr() is wrapper for userscript manager's XHR method, also requires chatgpt.js to get token):

    function speak() {
        const synthEndpoint = 'https://chatgpt.com/backend-api/synthesize'
        const msgID = '7fd31ce3-c6cc-4820-8166-4aed9f49f301'
        const convoID = '67695cab-a0f0-800f-b79f-481c030a9f87'
        const voice = 'cove'
        chatgpt.getAccessToken().then(token => {
            xhr({
                method: 'GET', url: `${synthEndpoint}?message_id=${msgID}&conversation_id=${convoID}&voice=${voice}`,
                headers: {
                    'Accept': '*/*', 'Accept-Encoding': 'gzip, deflate, br, zstd',
                    'Authorization': `Bearer ${token}`,
                    'Connection': 'keep-alive', 'DNT': '1',
                    'Host': 'chatgpt.com',
                    'OAI-Device-Id': 'b6e61267-b02e-42dc-837b-9e0d852af0e3', 'OAI-Language': 'en-US',
                    'Priority': 'u=4', 'Referer': 'https://chatgpt.com/c/67695cab-a0f0-800f-b79f-481c030a9f87',
                    'Sec-Fetch-Dest': 'empty', 'Sec-Fetch-Mode': 'cors', 'Sec-Fetch-Site': 'same-origin'
                },
                onload: resp => {
                    if (resp.status != 200) {
                        console.error('Response status', resp.status)
                        return console.info(resp)
                    }
                    console.log('Audio content received', resp)
                    const audioBlob = new Blob([resp.response], { type: 'audio/aac' }),
                          audioURL = URL.createObjectURL(audioBlob),
                          audio = new Audio(audioURL)
                    audio.play().then(() => console.log('Audio is playing'))
                        .catch(err => console.error('Error playing audio', err))
                }
            })
        })
    }

The msgID + convoID + voice + OAI-Device-Id + OAI-Language should be replaced with actual user values if anyone can get it to play (probably requires converting response data to AAC file) then I can make it dynamic for this script

[adamlui]

Actually upon reflection, I think the msg ID isn't generated until response completion but even if before, I think the msg thus audio file will not contain all the words yet so your feature is not possible

[happyf-weallareeuropean]

thank your respond adamlui! i was though use siri.
i not sure in rule of select / muti request to unreal real time. i well started try this tonight, share with you update later!

[adamlui]

@happyf-weallareeuropean are you really good at JS/APIs though? Because advanced knowledge is required to play with the code I pasted

[happyf-weallareeuropean]

nah, just start it swift 3 mo. from playgrounds switch to xcode as 2 week earlyer

[happyf-weallareeuropean]

i will take care, any stock share with u also!