Add ONNX export of facebook/mms-tts-pan for browser TTS.

README.md

@@ -0,0 +1,36 @@
+---
+language:
+- pa
+license: cc-by-nc-4.0
+library_name: transformers.js
+pipeline_tag: text-to-audio
+tags:
+- onnx
+- text-to-speech
+- tts
+- mms
+- facebook
+- pa
+- onnxruntime-web
+base_model: facebook/mms-tts-pan
+---
+
+# MMS-TTS Punjabi — ONNX
+
+Verbatim ONNX export of [`facebook/mms-tts-pan`](https://huggingface.co/facebook/mms-tts-pan)
+for in-browser inference via [transformers.js](https://huggingface.co/docs/transformers.js)
+or [onnxruntime-web](https://onnxruntime.ai/).
+
+This bundle was produced by the [onnx-conversion](https://github.com/NakliTechie/onnx-conversion) pipeline at
+`scripts/seq2seq/03_mms_batch_export.py` so the [Ghoshana](https://github.com/NakliTechie/Ghoshana)
+browser app can ship MMS-TTS coverage for Punjabi without depending on
+a system voice install.
+
+The MMS family from [Massively Multilingual Speech](https://research.facebook.com/publications/scaling-speech-technology-to-1000-languages/)
+covers 1107+ languages but only ~12 of those have community ONNX exports
+(Xenova mostly), with `kn` (Kannada) at onnx-community. This repo fills
+in the rest of the Punjabi gap.
+
+## License
+
+CC-BY-NC-4.0 (inherited from the upstream `facebook/mms-tts-pan`).

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "<unk>": 59
+}

config.json

@@ -0,0 +1,82 @@
+{
+  "activation_dropout": 0.1,
+  "architectures": [
+    "VitsModel"
+  ],
+  "attention_dropout": 0.1,
+  "depth_separable_channels": 2,
+  "depth_separable_num_layers": 3,
+  "dtype": "float32",
+  "duration_predictor_dropout": 0.5,
+  "duration_predictor_filter_channels": 256,
+  "duration_predictor_flow_bins": 10,
+  "duration_predictor_kernel_size": 3,
+  "duration_predictor_num_flows": 4,
+  "duration_predictor_tail_bound": 5.0,
+  "ffn_dim": 768,
+  "ffn_kernel_size": 3,
+  "flow_size": 192,
+  "hidden_act": "relu",
+  "hidden_dropout": 0.1,
+  "hidden_size": 192,
+  "initializer_range": 0.02,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "leaky_relu_slope": 0.1,
+  "model_type": "vits",
+  "noise_scale": 0.667,
+  "noise_scale_duration": 0.8,
+  "num_attention_heads": 2,
+  "num_hidden_layers": 6,
+  "num_speakers": 1,
+  "posterior_encoder_num_wavenet_layers": 16,
+  "prior_encoder_num_flows": 4,
+  "prior_encoder_num_wavenet_layers": 4,
+  "resblock_dilation_sizes": [
+    [
+      1,
+      3,
+      5
+    ],
+    [
+      1,
+      3,
+      5
+    ],
+    [
+      1,
+      3,
+      5
+    ]
+  ],
+  "resblock_kernel_sizes": [
+    3,
+    7,
+    11
+  ],
+  "sampling_rate": 16000,
+  "speaker_embedding_size": 0,
+  "speaking_rate": 1.0,
+  "spectrogram_bins": 513,
+  "transformers_version": "4.57.6",
+  "upsample_initial_channel": 512,
+  "upsample_kernel_sizes": [
+    16,
+    16,
+    4,
+    4
+  ],
+  "upsample_rates": [
+    8,
+    8,
+    2,
+    2
+  ],
+  "use_bias": true,
+  "use_stochastic_duration_prediction": true,
+  "vocab_size": 59,
+  "wavenet_dilation_rate": 1,
+  "wavenet_dropout": 0.0,
+  "wavenet_kernel_size": 5,
+  "window_size": 4
+}

model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b25586bf8b8549e101d59d661a0729d3c2670b3a2da317fabc13ce556e2deecc
+size 114302739

special_tokens_map.json

@@ -0,0 +1,4 @@
+{
+  "pad_token": "ਏ",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,31 @@
+{
+  "add_blank": true,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "ਏ",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "59": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "extra_special_tokens": {},
+  "is_uroman": false,
+  "language": "pan",
+  "model_max_length": 1000000000000000019884624838656,
+  "normalize": true,
+  "pad_token": "ਏ",
+  "phonemize": false,
+  "tokenizer_class": "VitsTokenizer",
+  "unk_token": "<unk>"
+}

vocab.json

@@ -0,0 +1,61 @@
+{
+  " ": 31,
+  "-": 5,
+  "6": 28,
+  "_": 7,
+  "l": 6,
+  "ਂ": 40,
+  "ਅ": 44,
+  "ਆ": 48,
+  "ਇ": 10,
+  "ਈ": 37,
+  "ਉ": 43,
+  "ਊ": 19,
+  "ਏ": 0,
+  "ਐ": 36,
+  "ਓ": 41,
+  "ਔ": 24,
+  "ਕ": 53,
+  "ਖ": 45,
+  "ਗ": 33,
+  "ਘ": 38,
+  "ਚ": 49,
+  "ਛ": 30,
+  "ਜ": 23,
+  "ਝ": 18,
+  "ਟ": 1,
+  "ਠ": 4,
+  "ਡ": 46,
+  "ਢ": 54,
+  "ਣ": 12,
+  "ਤ": 51,
+  "ਥ": 13,
+  "ਦ": 25,
+  "ਧ": 20,
+  "ਨ": 14,
+  "ਪ": 11,
+  "ਫ": 17,
+  "ਬ": 34,
+  "ਭ": 39,
+  "ਮ": 57,
+  "ਯ": 16,
+  "ਰ": 55,
+  "ਲ": 47,
+  "ਵ": 50,
+  "ਸ": 15,
+  "ਹ": 52,
+  "਼": 9,
+  "ਾ": 56,
+  "ਿ": 22,
+  "ੀ": 32,
+  "ੁ": 21,
+  "ੂ": 29,
+  "ੇ": 42,
+  "ੈ": 3,
+  "ੋ": 26,
+  "ੌ": 27,
+  "੍": 35,
+  "ੜ": 8,
+  "ੰ": 2,
+  "ੱ": 58
+}