Use facebook/mms-tts-eng instead of suno/bark in Gradio cookbook for Text2Speech #885
+24
−23
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Lol literally a 1-line fix. I tested by removing the model name explicitly from HF model parsers to use the default, but this errored since we're setting it to "" instead of None. <img width="610" alt="Screenshot 2024-01-11 at 03 02 00" src="https://github.com/lastmile-ai/aiconfig/assets/151060367/1a045dbb-ad1c-46db-ba1c-32b05d923484"> ## Test Plan Before ``` model_name="" ``` I got an error saying that no model "" exists, forgot to take screenshot but it was real pls believe me After ``` model_name=None No model was supplied, defaulted to sshleifer/distilbart-cnn-12-6 and revision a4f8f3e (https://huggingface.co/sshleifer/distilbart-cnn-12-6). Using a pipeline without specifying a model name and revision in production is not recommended. ``` https://github.com/lastmile-ai/aiconfig/assets/151060367/db6b35d4-9e03-4ff5-b035-c77263978fd4
See #877 for context, and #880 for why that original fix needed to be reverted Just a quick hack to get unblocked. I tried originally to make the ASR and Image2Text ParameterizedModel but that caused other errors. ## Test Plan Before https://github.com/lastmile-ai/aiconfig/assets/151060367/a23a5d5c-d9a2-415b-8a6e-9826da56e985 After https://github.com/lastmile-ai/aiconfig/assets/151060367/f29580e9-5cf6-43c5-b848-bb525eb368f2
…ms, get_output_text in right format Getting an error saying that "model" param is defined twice, which it is because we're passing it through completion params and again through the explicit model param. This along with the change in #879 will ensure that the right model of `None` is passed if the value is empty string. I also fixed typo in the aiconfig.json file ## Test Plan I also had to change the data field from `"./hi.mp3"` to `"cookbooks/Gradio/hi.mp3"` becuase I can only run the config from the high level `aiconfig` directory due to this line when running the editor: https://github.com/lastmile-ai/aiconfig/blob/b632094618c8f96b186b1849d0036e1313e02c5a/python/src/aiconfig/scripts/aiconfig_cli.py#L126 After https://github.com/lastmile-ai/aiconfig/assets/151060367/7387706f-deea-414b-b429-732c78322615
…Text2Speech Bark is ~5GB and took my laptop 30mins to download for the first time. The Facebook MMS (https://huggingface.co/facebook/mms-tts-eng?fbclid=IwAR22U5VSaZ_V7QW4TJe77CM7f4fsTHjqCMjgLKV31wUl66ww1vXYSrEl-MA) is only ~100MB which is way better and at least for cookbook I feel much more worth it to quickly test things out. Yes the output quality is way more memey and "voice AI" stereotype, but I feel the tradeoff is worth it
This was referenced Jan 11, 2024
rossdanlm
requested review from
saqadri,
rholinshead,
suyoglastmileai,
Ankush-lastmile and
jonathanlastmileai
as code owners
rossdanlm
added a commit
that referenced
this pull request
Jan 11, 2024
[ez][fix][HF]Explicitly set hf_model to None if it's empty string Lol literally a 1-line fix. I tested by removing the model name explicitly from HF model parsers to use the default, but this errored since we're setting it to "" instead of None. <img width="610" alt="Screenshot 2024-01-11 at 03 02 00" src="https://github.com/lastmile-ai/aiconfig/assets/151060367/1a045dbb-ad1c-46db-ba1c-32b05d923484"> ## Test Plan Before ``` model_name="" ``` I got an error saying that no model "" exists, forgot to take screenshot but it was real pls believe me After ``` model_name=None No model was supplied, defaulted to sshleifer/distilbart-cnn-12-6 and revision a4f8f3e (https://huggingface.co/sshleifer/distilbart-cnn-12-6). Using a pipeline without specifying a model name and revision in production is not recommended. ``` https://github.com/lastmile-ai/aiconfig/assets/151060367/db6b35d4-9e03-4ff5-b035-c77263978fd4 --- Stack created with [Sapling](https://sapling-scm.com). Best reviewed with [ReviewStack](https://reviewstack.dev/lastmile-ai/aiconfig/pull/879). * #885 * #883 * #882 * #881 * __->__ #879
saqadri
added a commit
that referenced
this pull request
Jan 11, 2024
#881) Revert "[HF][fix] Use ParameterizedModelParser instead of ModelParser" Reverts #877 This caused a bug because image and audio inputs can't be parametrized. See #882 for a better fix instead --- Stack created with [Sapling](https://sapling-scm.com). Best reviewed with [ReviewStack](https://reviewstack.dev/lastmile-ai/aiconfig/pull/881). * #885 * #883 * #882 * __->__ #881 * #879
saqadri
added a commit
that referenced
this pull request
Jan 11, 2024
[HF][fix] Allow `kwargs` in `ModelParser.run()` See #877 for context, and #880 for why that original fix needed to be reverted Just a quick hack to get unblocked. I tried originally to make the ASR and Image2Text ParameterizedModel but that caused other errors. ## Test Plan Before https://github.com/lastmile-ai/aiconfig/assets/151060367/a23a5d5c-d9a2-415b-8a6e-9826da56e985 After https://github.com/lastmile-ai/aiconfig/assets/151060367/f29580e9-5cf6-43c5-b848-bb525eb368f2 --- Stack created with [Sapling](https://sapling-scm.com). Best reviewed with [ReviewStack](https://reviewstack.dev/lastmile-ai/aiconfig/pull/882). * #885 * #883 * __->__ #882 * #881 * #879
saqadri
added a commit
that referenced
this pull request
Jan 11, 2024
…ms, get_output_text in right format (#883) [HF][fix] Small nit fixes to ASR: Remove "model" from completion params, get_output_text in right format Getting an error saying that "model" param is defined twice, which it is because we're passing it through completion params and again through the explicit model param. This along with the change in #879 will ensure that the right model of `None` is passed if the value is empty string. I also fixed typo in the aiconfig.json file ## Test Plan I also had to change the data field from `"./hi.mp3"` to `"cookbooks/Gradio/hi.mp3"` becuase I can only run the config from the high level `aiconfig` directory due to this line when running the editor: https://github.com/lastmile-ai/aiconfig/blob/b632094618c8f96b186b1849d0036e1313e02c5a/python/src/aiconfig/scripts/aiconfig_cli.py#L126 After https://github.com/lastmile-ai/aiconfig/assets/151060367/7387706f-deea-414b-b429-732c78322615 --- Stack created with [Sapling](https://sapling-scm.com). Best reviewed with [ReviewStack](https://reviewstack.dev/lastmile-ai/aiconfig/pull/883). * #885 * __->__ #883 * #882 * #881 * #879
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Use facebook/mms-tts-eng instead of suno/bark in Gradio cookbook for Text2Speech
Bark is ~5GB and took my laptop 30mins to download for the first time. The Facebook MMS (https://huggingface.co/facebook/mms-tts-eng?fbclid=IwAR22U5VSaZ_V7QW4TJe77CM7f4fsTHjqCMjgLKV31wUl66ww1vXYSrEl-MA) is only ~100MB which is way better and at least for cookbook I feel much more worth it to quickly test things out. Yes the output quality is way more memey and "voice AI" stereotype, but I feel the tradeoff is worth it
Stack created with Sapling. Best reviewed with ReviewStack.
kwargsinModelParser.run()#882