holy shit that worked

tts_service/models.toml

@@ -11,7 +11,8 @@ melgan = "melgan_trump_txlearn_2020-05-30_steno_chkpt9800.torchjit"
 name = "Donald Trump (Glow-TTS experiments)"
 slug = "trump2"
 model_pipeline = "ArpabetGlowTtsMelgan"
-glow_tts = "glow-experimental.torchjit"
+#glow_tts = "glow-experimental.torchjit"
+glow_tts = "output.torchjit"
 melgan = "melgan_trump_txlearn_2020-05-30_steno_chkpt9800.torchjit"
 
 [[speakers]]
@@ -29,7 +30,8 @@ base_directory = "/home/bt/dev/voder/tacotron_melgan"
 
 [[model_locations]]
 model_type = "ArpabetGlowTts" # TODO: This is not the Arpabet I'm using
-base_directory = "/home/bt/dev/jit-models/output/glow-tts-jaywalnut"
+#base_directory = "/home/bt/dev/jit-models/output/glow-tts-jaywalnut"
+base_directory = "/home/bt/dev/jit-models/glow-tts-jaywalnut"
 
 [[model_locations]]
 model_type = "Melgan"

tts_service/src/model/arpabet_glow_tts_model.rs

@@ -1,5 +1,5 @@
 use crate::model::model_container::ModelContainer;
-use anyhow::Result;
+use anyhow::{Result, Error};
 use std::path::Path;
 use std::thread;
 use std::time::Duration;
@@ -24,6 +24,68 @@ impl ArpabetGlowTtsModel {
     let text_tensor = text_tensor.unsqueeze(0);
     println!("Text tensor unsq: {:?}", text_tensor);
 
-    self.model_container.forward(&text_tensor, )
+    let lengths_tensor = Tensor::of_slice(&[text_buffer.len() as i64]);
+
+    /*
+
+    "this is the song that never ends"
+
+    x_tst tensor([[ 91, 109, 131,  11, 109, 146,  11,  91,  73,  11, 131,  78, 120,  11,
+          91,  70, 133,  11, 119,  94, 143,  97,  11,  94, 119,  90, 146]],
+       device='cuda:0')
+
+    x_tst_lengths tensor([27], device='cuda:0')
+
+     */
+
+    /*
+    r: Expected tensor for argument #1 \'indices\' to have scalar type Long;
+    but got CPUIntType instead (while checking arguments for embedding)\n" }',
+    tts_service/src/model/model_container.rs:33:5
+
+    r: Expected tensor for argument #1 \'indices\' to have scalar type Long;
+    but got CPUFloatType instead (while checking arguments for embedding)\n" }',
+    tts_service/src/model/model_container.rs:33:5
+     */
+
+    /*
+
+    tst_stn 39  testing this text to speech generator
+    sequence 1 [[ 11 133  94 131 133 108 120  11  91 109 131  11 133  94 116 131 133  11
+                  133 141  11 131 129 113  89  11 115  94 119  97 103 133  97  11]]
+
+    x_tst 1 tensor([[ 11, 133,  94, 131, 133, 108, 120,  11,  91, 109, 131,  11, 133,  94,
+         116, 131, 133,  11, 133, 141,  11, 131, 129, 113,  89,  11, 115,  94,
+         119,  97, 103, 133,  97,  11]])
+
+   x_tst_lengths 1 tensor([34])
+
+     */
+   // let text : Vec<i64> = vec![ 91, 109, 131,  11, 109, 146,  11,  91,  73,  11, 131,  78, 120,  11,
+   //   91,  70, 133,  11, 119,  94, 143,  97,  11,  94, 119,  90, 146];
+
+    let text : Vec<i64> = vec![11, 133, 94, 131, 133, 108, 120, 11, 91, 109, 131,
+      11, 133, 94, 116, 131, 133, 11, 133, 141, 11, 131, 129, 113, 89, 11, 115, 94,
+      119, 97, 103, 133, 97, 11];
+
+
+    //let text : Vec<f32> = text.iter().map(|n| *n as i64).collect();
+
+    //let one_d : Vec<&[i64]> = vec![text.as_slice()];
+    //let two_d : &[&[i64]] = one_d.as_slice();
+
+    let x_tst = Tensor::of_slice(&text.as_slice());
+
+    println!("\n\nTensor size: {:?}", x_tst.size());
+
+    let x_tst_2 = x_tst.unsqueeze(0);
+
+    println!("\n\nTensor unsqueezed size: {:?}", x_tst_2.size());
+
+    //let lengths = [27i64];
+    let lengths = [34i64];
+    let x_tst_lengths = Tensor::of_slice(&lengths);
+
+    self.model_container.forward2(&x_tst_2, &x_tst_lengths)
   }
 }

tts_service/src/model/model_container.rs

@@ -3,6 +3,8 @@ use tch::CModule;
 use tch::Tensor;
 use tch::nn::Module;
 use std::path::{Path, PathBuf};
+use anyhow::Result as AnyhowResult;
+use std::fmt::Formatter;
 
 /// Holds the loaded pytorch JIT model
 pub struct ModelContainer {
@@ -28,4 +30,14 @@ impl ModelContainer {
   pub fn forward(&self, tensor: &Tensor) -> Tensor {
     self.jit_model.forward(tensor)
   }
+
+  pub fn forward2(&self, arg1: &Tensor, arg2: &Tensor) -> Tensor {
+    let result = self.jit_model.forward_ts(&[arg1, arg2]);
+
+    if let Err(err) = result.as_ref() {
+      println!("error: {:?}", err);
+    }
+
+    result.expect("SHOULD WORK")
+  }
 }