Add Xav.Reader.stream!/2

README.md

@@ -59,9 +59,6 @@ Kino.Image.new(tensor)
 Speech to text:
 
 ```elixir
-# See https://hexdocs.pm/bumblebee/Bumblebee.Audio.WhisperFeaturizer.html for default sampling rate
-r = Xav.Reader.new!("sample.mp3", read: :audio, out_format: :f32, out_channels: 1, out_sample_rate: 16_000)
-
 {:ok, whisper} = Bumblebee.load_model({:hf, "openai/whisper-tiny"})
 {:ok, featurizer} = Bumblebee.load_featurizer({:hf, "openai/whisper-tiny"})
 {:ok, tokenizer} = Bumblebee.load_tokenizer({:hf, "openai/whisper-tiny"})
@@ -72,12 +69,12 @@ serving =
     defn_options: [compiler: EXLA]
   )
 
-# read a couple of frames
+# Read a couple of frames.
+# See https://hexdocs.pm/bumblebee/Bumblebee.Audio.WhisperFeaturizer.html for default sampling rate.
 frames =
-  for _i <- 0..200 do
-    {:ok, frame} = Xav.Reader.next_frame(r)
-    Xav.Frame.to_nx(frame)
-  end
+    Xav.Reader.stream!("sample.mp3", read: :audio, out_format: :f32, out_channels: 1, out_sample_rate: 16_000)
+    |> Stream.take(200)
+    |> Enum.map(fn frame -> Xav.Reader.to_nx(frame) end)
 
 batch = Nx.Batch.concatenate(frames)
 batch = Nx.Defn.jit_apply(&Function.identity/1, [batch])

lib/reader.ex

@@ -129,6 +129,31 @@ defmodule Xav.Reader do
     end
   end
 
+  @doc """
+  Creates a new reader stream.
+  """
+  @spec stream!(String.t(), opts()) :: Enumerable.t()
+  def stream!(path, opts \\ []) do
+    Stream.resource(
+      fn ->
+        case new(path, opts) do
+          {:ok, reader} ->
+            reader
+
+          {:error, reason} ->
+            raise "Couldn't create a new Xav.Reader stream. Reason: #{inspect(reason)}"
+        end
+      end,
+      fn reader ->
+        case next_frame(reader) do
+          {:ok, frame} -> {[frame], reader}
+          {:error, :eof} -> {:halt, reader}
+        end
+      end,
+      fn _reader -> :ok end
+    )
+  end
+
   defp to_human_readable(:libdav1d), do: :av1
   defp to_human_readable(:mp3float), do: :mp3
   defp to_human_readable(other), do: other

test/reader_test.exs

@@ -17,6 +17,11 @@ defmodule Xav.ReaderTest do
     for _i <- 0..(30 * 5), do: assert({:ok, %Xav.Frame{}} = Xav.Reader.next_frame(r))
   end
 
+  test "stream!" do
+    Xav.Reader.stream!("./test/fixtures/sample_h264.mp4")
+    |> Enum.all?(fn frame -> is_struct(frame, Xav.Frame) end)
+  end
+
   test "to_nx/1" do
     {:ok, r} = Xav.Reader.new("./test/fixtures/sample_h264.mp4")
     {:ok, frame} = Xav.Reader.next_frame(r)
@@ -70,14 +75,6 @@ defmodule Xav.ReaderTest do
   end
 
   defp test_speech_to_text(path, expected_output) do
-    reader =
-      Xav.Reader.new!(path,
-        read: :audio,
-        out_channels: 1,
-        out_format: :f32,
-        out_sample_rate: 16_000
-      )
-
     {:ok, whisper} = Bumblebee.load_model({:hf, "openai/whisper-tiny"})
     {:ok, featurizer} = Bumblebee.load_featurizer({:hf, "openai/whisper-tiny"})
     {:ok, tokenizer} = Bumblebee.load_tokenizer({:hf, "openai/whisper-tiny"})
@@ -89,7 +86,12 @@ defmodule Xav.ReaderTest do
       )
 
     batch =
-      read_frames(reader)
+      Xav.Reader.stream!(path,
+        read: :audio,
+        out_channels: 1,
+        out_format: :f32,
+        out_sample_rate: 16_000
+      )
       |> Enum.map(&Xav.Frame.to_nx(&1))
       |> Nx.Batch.concatenate()
 
@@ -98,14 +100,4 @@ defmodule Xav.ReaderTest do
 
     assert [%{text: ^expected_output}] = chunks
   end
-
-  defp read_frames(reader, acc \\ []) do
-    case Xav.Reader.next_frame(reader) do
-      {:ok, frame} ->
-        read_frames(reader, [frame | acc])
-
-      {:error, :eof} ->
-        Enum.reverse(acc)
-    end
-  end
 end