I uploaded great sample for the voice test and it tells me this...

[binnar22]

When cloning a voice, I decided to search for the best sample I could find. For example, I downloaded Mr Fantastic ( from Fallout New vegas ) for voice sample and after imported it in and wanting to see the result, it said this :

" Automatic reference transcription needs an installed speech-to-text model. Provide a matching reference transcript, or install and select a speech-to-text model in Model Catalogue, then try again. "

I'm still new to this so I don't know how to solve it. Can anyone help me pl;s?

[tuht97]

Trim your reference audio file so it is under 15-20 seconds long.

[Jeremy-Jud]

Provide a reference transcript (fast)
Click "Import script" or paste the exact text that is spoken in the audio into the Script box.
Make sure the language setting matches the audio.
Try the voice test again.
Install and select a speech‑to‑text model (auto transcription)
Open Model Catalogue (or Settings → Model Catalogue).
Find a speech‑to‑text model (e.g. a Whisper/STT model) and install it.
Select that speech‑to‑text model for Automatic reference transcription.
Re-run the voice cloning test — the app will transcribe the audio automatically.
Notes

The transcript must match the audio fairly closely for good results.
If the audio is from copyrighted material (game/film), make sure you have the right to use it.

[binnar22]

Thank you for replying to my email ^^ I have gone back to check the settings you mentioned. ( The Whisper model ) How can I select that for Automatic reference subscription? I tried to go back and generate the voiceline but it doesn't work Vào Thứ 5, 15 thg 10, 2026 vào lúc 14:21 Jeremy-Jud < ***@***.***> đã viết:

…

Provide a reference transcript (fast) Click "Import script" or paste the exact text that is spoken in the audio into the Script box. Make sure the language setting matches the audio. Try the voice test again. Install and select a speech‑to‑text model (auto transcription) Open Model Catalogue (or Settings → Model Catalogue). Find a speech‑to‑text model (e.g. a Whisper/STT model) and install it. Select that speech‑to‑text model for Automatic reference transcription. Re-run the voice cloning test — the app will transcribe the audio automatically. Notes The transcript must match the audio fairly closely for good results. If the audio is from copyrighted material (game/film), make sure you have the right to use it. — Reply to this email directly, view it on GitHub <#2604?email_source=notifications&email_token=CQEMJ6K5L23O7IHF45L5FX35UB3GPA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQZTGNZVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-18943375>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CQEMJ6KMKVJJCBHWTAYOF7T5UB3GPAVCNFSNUABJKJSXA33TNF2G64TZHMYTEMBWGM4TANJXGE5UI2LTMN2XG43JN5XDWMJQHE2DQNZSHCQXMAQ> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/CQEMJ6KHMEDQ4M2FN2ME2X35UB3GPA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQZTGNZVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVJTG633UMVZF62LPOM> and Android <https://github.com/notifications/mobile/android/CQEMJ6MZ67RVD7ZXDF3PMRT5UB3GPA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQZTGNZVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVZTG633UMVZF6YLOMRZG62LE>. Download it today! You are receiving this because you authored the thread.Message ID: ***@***.***>

[Jeremy-Jud]

Hi @binnar22 — thanks for the follow-up! To select Whisper for automatic reference transcription, do this:

Open Settings → Model Catalogue (or click Model Catalogue from the app).
Search for a speech‑to‑text / Whisper model (names vary: “Whisper”, “whisper-small/medium/large”, or “STT/Whisper”).
Click Install and wait until the model shows as Installed/Downloaded.
Open the Voice Test dialog (or Settings → Automatic reference transcription) and find the “Automatic reference transcription” dropdown — select the Whisper/STT model you just installed.
Make sure the Script box is empty (so the app will transcribe) or click “Import script” / paste a matching transcript if you already have one.
Confirm the language setting matches the audio, then re-run the voice test.
If it still doesn’t work:
contact a professional

[binnar22]

Okay so I have tried something else and noticed something. It's the length of the sample. It said that it can have a maximum of 30 seconds of voice sample. So I imported a 25 seconds+ long of voice for sample. It couldn't register all of that. I then changed to about 10 seconds+ and now the voice is fine. Thanks for your help Vào Thứ 5, 15 thg 10, 2026 vào lúc 15:31 Jeremy-Jud < ***@***.***> đã viết:

…

Hi @binnar22 <https://github.com/binnar22> — thanks for the follow-up! To select Whisper for automatic reference transcription, do this: Open Settings → Model Catalogue (or click Model Catalogue from the app). Search for a speech‑to‑text / Whisper model (names vary: “Whisper”, “whisper-small/medium/large”, or “STT/Whisper”). Click Install and wait until the model shows as Installed/Downloaded. Open the Voice Test dialog (or Settings → Automatic reference transcription) and find the “Automatic reference transcription” dropdown — select the Whisper/STT model you just installed. Make sure the Script box is empty (so the app will transcribe) or click “Import script” / paste a matching transcript if you already have one. Confirm the language setting matches the audio, then re-run the voice test. If it still doesn’t work: contact a professional — Reply to this email directly, view it on GitHub <#2604?email_source=notifications&email_token=CQEMJ6OP65L7P2EHVF73GKD5UCDMVA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQ2DMMZSUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-18944632>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CQEMJ6PCZYY6AW74FSTRWPD5UCDMVAVCNFSNUABJKJSXA33TNF2G64TZHMYTEMBWGM4TANJXGE5UI2LTMN2XG43JN5XDWMJQHE2DQNZSHCQXMAQ> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/CQEMJ6MPADM7J3LINNPPRV35UCDMVA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQ2DMMZSUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSVGM33PORSXEX3JN5ZQ> and Android <https://github.com/notifications/mobile/android/CQEMJ6JLVYQMOAI2DYEKGPD5UCDMVA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBZGQ2DMMZSUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>. Download it today! You are receiving this because you were mentioned.Message ID: ***@***.***>

[binnar22]

thank you. It worked Vào Thứ 4, 7 thg 10, 2026 vào lúc 11:44 Tu Hinh ***@***.***> đã viết:

…

Trim your reference audio file so it is under 15-20 seconds long. — Reply to this email directly, view it on GitHub <#2604?email_source=notifications&email_token=CQEMJ6J6SB4ID2WYCLHRVPD5SXC2LA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBXHA4DMMRXUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-18788627>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CQEMJ6I4FI73IVV4ONSWTJT5SXC2LAVCNFSNUABJKJSXA33TNF2G64TZHMYTEMBWGM4TANJXGE5UI2LTMN2XG43JN5XDWMJQHE2DQNZSHCQXMAQ> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/CQEMJ6IE2XPALWKKOOKY2DL5SXC2LA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBXHA4DMMRXUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVJTG633UMVZF62LPOM> and Android <https://github.com/notifications/mobile/android/CQEMJ6KNYI7FD5G2YSXQYVD5SXC2LA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBXHA4DMMRXUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVZTG633UMVZF6YLOMRZG62LE>. Download it today! You are receiving this because you authored the thread.Message ID: ***@***.***>

[Dai-Katana]

I see. So it was length for me also. While clicking around somewhere it said up to 75sec even tho in cloning section says 5-15sec. Tried first with 75 sec and this error, with 15 sec works well.