Trying to Transcribe but getting error: 'Missing dataset: ./training/voice1//whisper.json' #223

New Issue

embanot · 2023-04-30T18:53:00Z

embanot commented

2023-04-30 18:53:00 +00:00

I'm trying to prepare my dataset for training. I put in all the wav files into a folder under the Voices directory. I've left all the settings in Prepare Dataset tab as default. I select the dataset source to be the folder I put all my wavs into. Then I hit Transcribe and Process. The console output gives me this:

Processed dataset to: ./training/voice1/
Missing dataset: ./training/voice1//whisper.json

Also when I look at the command prompt, it's failing to transcribe each of my wav files giving an error saying that the system cannot find the file specified. But all the wav files are there so I dunno why its giving me this message.

Any idea why this is happening?

I'm trying to prepare my dataset for training. I put in all the wav files into a folder under the Voices directory. I've left all the settings in Prepare Dataset tab as default. I select the dataset source to be the folder I put all my wavs into. Then I hit Transcribe and Process. The console output gives me this: Processed dataset to: ./training/voice1/ Missing dataset: ./training/voice1//whisper.json Also when I look at the command prompt, it's failing to transcribe each of my wav files giving an error saying that the system cannot find the file specified. But all the wav files are there so I dunno why its giving me this message. Any idea why this is happening?

pheonis commented

2023-05-01 07:56:19 +00:00

This thread might solve this #177

This thread might solve this https://git.ecker.tech/mrq/ai-voice-cloning/issues/177

nanonomad commented

2023-05-01 18:45:38 +00:00

I think I fixed this with:
cd ai-voice-cloning
source ./venv/bin/activate
sudo apt install ffmpeg (and maybe a pip install ffmpeg)
deactivate

I think I fixed this with: cd ai-voice-cloning source ./venv/bin/activate sudo apt install ffmpeg (and maybe a pip install ffmpeg) deactivate

👍 1

mrq commented

2023-05-03 22:35:26 +00:00

I might have incidentally fixed this in commit 853c7fdccf, if you were using "Transcribe All" instead (which transcribes ALL voices, rather than the selected one).

In short, my training rig had the transcribe and slice blocks commented out for transcribing all and it somehow made its way upstream. Apologies.

I might have incidentally fixed this in commit 853c7fdccf13277e500307aa241e149d7e99ac7d, if you were using "Transcribe All" instead (which transcribes ALL voices, rather than the selected one). In short, my training rig had the transcribe and slice blocks commented out for transcribing all and it somehow made its way upstream. Apologies.

Sign in to join this conversation.