{"owner":"flashlight","repo":"wav2letter","hasSkills":true,"totalSkillsCount":1,"totalTokensCount":354,"categories":["plugin-manifest"],"hasMcp":false,"mcpConfig":null,"found":["recipes/streaming_convnets/tools/README.md"],"skills":{"recipes/streaming_convnets/tools/README.md":"# Tools\n\n## Streaming TDS model conversion for running inference pipeline\nOnce a model is trained in wav2letter++ for streaming TDS models using the [provided recipe](https://github.com/flashlight/wav2letter/tree/master/recipes/models/streaming_convnets) possibly customized to suit ones' use-case, the model needs to be serialized to a format which wav2letter@anywhere inference platform can load. `StreamingTDSModelConverter` can be used to do this. Note that the script only supports models trained using the streaming TDS + CTC style architectures as described in the paper [here](https://research.fb.com/publications/scaling-up-online-speech-recognition-using-convnets/).\n### Using the Pipeline\nBuild the tool with `make streaming_tds_model_converter`.\nAnd to run the binary:\n```\n[path to binary]/streaming_tds_model_converter \\\n    -am [path to model] \\\n    --outdir [output directory]\n```\nThe output directory will contain\n- `tokens.txt` - Tokens file (with blank symbol included)\n- `acoustic_model.bin` - Serialized acoutic model\n- `feature_extractor.bin` - Serialized feature extraction model which perform log-mel feature extraction and local normalization\n\nThese files can be used to run inference on audio files along with a few other files required for decoding like language model, lexicon etc. See the [tutorial](https://github.com/flashlight/wav2letter/wiki/Inference-Run-Examples) for more details.\n"},"files":{"recipes/streaming_convnets/tools/README.md":"# Tools\n\n## Streaming TDS model conversion for running inference pipeline\nOnce a model is trained in wav2letter++ for streaming TDS models using the [provided recipe](https://github.com/flashlight/wav2letter/tree/master/recipes/models/streaming_convnets) possibly customized to suit ones' use-case, the model needs to be serialized to a format which wav2letter@anywhere inference platform can load. `StreamingTDSModelConverter` can be used to do this. Note that the script only supports models trained using the streaming TDS + CTC style architectures as described in the paper [here](https://research.fb.com/publications/scaling-up-online-speech-recognition-using-convnets/).\n### Using the Pipeline\nBuild the tool with `make streaming_tds_model_converter`.\nAnd to run the binary:\n```\n[path to binary]/streaming_tds_model_converter \\\n    -am [path to model] \\\n    --outdir [output directory]\n```\nThe output directory will contain\n- `tokens.txt` - Tokens file (with blank symbol included)\n- `acoustic_model.bin` - Serialized acoutic model\n- `feature_extractor.bin` - Serialized feature extraction model which perform log-mel feature extraction and local normalization\n\nThese files can be used to run inference on audio files along with a few other files required for decoding like language model, lexicon etc. See the [tutorial](https://github.com/flashlight/wav2letter/wiki/Inference-Run-Examples) for more details.\n"},"items":[{"name":"README.md","path":"recipes/streaming_convnets/tools/README.md","rawUrl":"https://raw.githubusercontent.com/flashlight/wav2letter/HEAD/recipes/streaming_convnets/tools/README.md","title":"tools Documentation","category":"plugin-manifest","format":"markdown","content":"# Tools\n\n## Streaming TDS model conversion for running inference pipeline\nOnce a model is trained in wav2letter++ for streaming TDS models using the [provided recipe](https://github.com/flashlight/wav2letter/tree/master/recipes/models/streaming_convnets) possibly customized to suit ones' use-case, the model needs to be serialized to a format which wav2letter@anywhere inference platform can load. `StreamingTDSModelConverter` can be used to do this. Note that the script only supports models trained using the streaming TDS + CTC style architectures as described in the paper [here](https://research.fb.com/publications/scaling-up-online-speech-recognition-using-convnets/).\n### Using the Pipeline\nBuild the tool with `make streaming_tds_model_converter`.\nAnd to run the binary:\n```\n[path to binary]/streaming_tds_model_converter \\\n    -am [path to model] \\\n    --outdir [output directory]\n```\nThe output directory will contain\n- `tokens.txt` - Tokens file (with blank symbol included)\n- `acoustic_model.bin` - Serialized acoutic model\n- `feature_extractor.bin` - Serialized feature extraction model which perform log-mel feature extraction and local normalization\n\nThese files can be used to run inference on audio files along with a few other files required for decoding like language model, lexicon etc. See the [tutorial](https://github.com/flashlight/wav2letter/wiki/Inference-Run-Examples) for more details.\n","isInternal":false,"tokens":354,"sizeBytes":1416}],"systemPromptSnippet":"<agent_rules repository=\"flashlight/wav2letter\">\n\n<!-- Skill/Rule: tools Documentation (recipes/streaming_convnets/tools/README.md) -->\n# Tools\n\n## Streaming TDS model conversion for running inference pipeline\nOnce a model is trained in wav2letter++ for streaming TDS models using the [provided recipe](https://github.com/flashlight/wav2letter/tree/master/recipes/models/streaming_convnets) possibly customized to suit ones' use-case, the model needs to be serialized to a format which wav2letter@anywhere inference platform can load. `StreamingTDSModelConverter` can be used to do this. Note that the script only supports models trained using the streaming TDS + CTC style architectures as described in the paper [here](https://research.fb.com/publications/scaling-up-online-speech-recognition-using-convnets/).\n### Using the Pipeline\nBuild the tool with `make streaming_tds_model_converter`.\nAnd to run the binary:\n```\n[path to binary]/streaming_tds_model_converter \\\n    -am [path to model] \\\n    --outdir [output directory]\n```\nThe output directory will contain\n- `tokens.txt` - Tokens file (with blank symbol included)\n- `acoustic_model.bin` - Serialized acoutic model\n- `feature_extractor.bin` - Serialized feature extraction model which perform log-mel feature extraction and local normalization\n\nThese files can be used to run inference on audio files along with a few other files required for decoding like language model, lexicon etc. See the [tutorial](https://github.com/flashlight/wav2letter/wiki/Inference-Run-Examples) for more details.\n\n\n</agent_rules>"}